G-CART Autonomous Navigation and Controlsedge.rit.edu/content/OldEDGE/public/Archives/P05107/... ·...
Transcript of G-CART Autonomous Navigation and Controlsedge.rit.edu/content/OldEDGE/public/Archives/P05107/... ·...
Project 05107
Darren Rowen
Scott Glover
SatSat Fox
Chaichat Boonyarat
Delwin Guiao
Derick Call
Luan Nguyen
1 RECOGNIZE AND QUANTIFY NEEDS............................................................................41.1 Mission Statement...........................................................................................................41.2 Project Description..........................................................................................................41.3 Scope Limitations............................................................................................................51.4 Stakeholders.....................................................................................................................71.5 Key Business Goals.........................................................................................................71.6 Primary Market................................................................................................................71.7 Secondary Market............................................................................................................81.8 Innovation Opportunities.................................................................................................81.9 Background Research......................................................................................................81.10 Formal Statement of Work..............................................................................................9
2 CONCEPT DEVELOPMENT............................................................................................102.1 Subgroup........................................................................................................................112.2 Bus Architecture Concepts............................................................................................132.3 Steering Concepts..........................................................................................................182.4 Velocity Control Concepts............................................................................................232.5 Emergency Brake Concepts...........................................................................................31
3 FEASIBILITY........................................................................................................................353.1 Bus Architecture Feasibility..........................................................................................353.2 Steering Feasibility........................................................................................................373.3 Speed Control Feasibility..............................................................................................403.4 Feasibility Conclusion...................................................................................................43
4 OBJECTIVES & SPECIFICATIONS................................................................................444.1 Design Objectives.......................................................................................................444.2 Performance Specifications.......................................................................................454.3 Safety Issues...............................................................................................................47
5 DESIGN ANALYSIS & SIMULATION..............................................................................475.1 Steering Design and Simulation....................................................................................475.2 Speed Control Design and Simulation...........................................................................60
6 FUTURE PLANS.................................................................................................................656.1 Schedule.........................................................................................................................656.2 Budget............................................................................................................................66
7 CONCLUSION.....................................................................................................................668 REFERENCES....................................................................................................................679 APPENDIX...........................................................................................................................70
9.1 Steering control PIC Program Flow chart.....................................................................70
1 RECOGNIZE AND QUANTIFY NEEDS1.1 Mission Statement
The purpose of this senior design team is to improve on the current G-Cart design already
developed at the Rochester Institute of Technology (RIT). The new design will allow for the
vehicle to autonomously navigate a G.P.S. bound all-terrain course.
1.2 Project Description
Complete a 300 km all terrain course in less than 10 hours.
Have to avoid a variety of obstacles while staying with the G.P.S boundaries
Output Controls:
A - Steering right/left
B - Vehicle stop/start/idle
C - Speed and gear position (drive/neutral/reverse)
D – Brakes, master cylinder pressure 0-100 BARR
E - Siren and Sound Alert (on/off combinations). There are vehicle requirements in the
Grand Challenge rules. The expectation is safety assurance.
F – Throttle position for speed control
G – Power and speed override, to shut the vehicle down as fast as possible.
Sensor Inputs:
A – Location. G.P.S. course guidelines will be stored in the navigation control
computer
B – Road recognition “visual location”. Multiple environment mapping systems will be
used as input to the control systems
C – Kill switch (if any problems arise)
D – Camera for troubleshooting only. This camera can be strategically placed to watch a
potentially failing mechanical system.
E - Brake by wire input (value 0-100 BARR)
F –Throttle sensor input (velocity variation 0-255), incremental values spanning the
vehicles desired velocity capability (i.e. -5 to 60 mph)
Mechanical:
Steering control
Sensor integrity (ex. shock absorbing, cooling system, power supply)
Continental Teves prototype brake by wire system.
Emergency brake system.
Exterior manually activated kill switch.
The vehicle has to comply with so many rules that it has become part of the product description.
The rule book is many pages that will be attached to the report and located in the senior design
team notebook.
1.3 Scope Limitations
Funding is expected, but not yet available. Due to the complexity of this project, funding
the best solution could become a problem that affects the end results (it is a race; finishing and
winning may be two different things)
Due to competition deadlines and expected delivery time of actual race vehicle, testing
could be limited (for our team, not the project as a whole). Also, the new G-CART vehicle has
not yet been acquired (as of November 12, 2004), which is a major limiting factor of our design.
Our team is currently waiting on the many expected donations. The Executive Board as
well as our team mentor believes that almost everything will be donated. While this is probably
true, our team has to complete our portion of the design and implementation by the end February.
We have placed numerous requests for donations, but we may not have time to wait on the
companies’ decision processes.
Determining what is expected of our design team. The Executive Boards needs have been
hazy at best. We have redefined our responsibilities several times, based on the Executive
Boards input. Our teams reduced role in the overall project is proving to be problematic with
respect to the design of deliverable goods to satisfy our senior design project. We are currently
broken into two sub teams responsible for steering of the vehicle and speed control. The
objective is to resolve the steering and speed control as soon as possible, and move on to the
navigation system or any other subsystem that may require added support.
Another limitation is our team’s lack of VHDL knowledge and time to learn it, for the
purpose of FPGAs. Some of the members are currently exploring the feasibility of an FPGA
solution. Further information is contained in the FPGA design section.
1.4 Stakeholders
R.I.T could win the two million dollar prize money.
D.A.R.P.A is the host of the race (military)
Automotive industry could use some or all of the design
Public transportation (bus, taxi)
Continental Teves
Evolution Robotics, Inc.
Egnite Software
Nation Instrument
Vehicle manufacturer
EE Faculty advisors
1.5 Key Business Goals
Create funding for the completion of the project
Stay within budget, which minimal in comparison to the expected funding
Win the two million dollars prize for a first place finish
Advertise our sponsors that have donated money or products that aided the design and
completion of the autonomous vehicle.
Advertise Rochester Institute of Technology. Hopefully in a beneficial manner.
1.6 Primary Market
D.A.R.P.A is the host of the race (military)
1.7 Secondary Market
Car manufactures – autopilot function or advanced cruise control system
RIT research and development for EE/ME/CE, which may attract R&D funds
1.8 Innovation Opportunities
High speed navigation
Save human life (military vehicle with no operator or passengers on board),
Artificial Intelligent vehicle
Optimize for less error in comparison to human operation (commercial vehicle with
passengers)
1.9 Background Research
The purpose of the G-Cart team is to design a vehicle capable of autonomous navigation
through a variety of terrain. Once the autonomous vehicle is developed, it is the goal of the team
to compete in various competitions throughout the United States. The venues for competition
include DARPA’s Grande Challenge; a race of autonomous ground vehicles from Los Angeles
to Las Vegas with a cash award of two million dollars to the team that completes the course in
the shortest time.
To accomplish this goal, G-Cart plans to research the existing technology in the generic
field of autonomous vehicular control and apply our knowledge to further research being
conducted in this area by RIT faculty, staff, and senior design teams. Ultimately, after our initial
goal of constructing an autonomous vehicle has been reached, we will continue research and
development in the realm of autonomous navigation and control. It is our long term hope to
develop a streamline system of autonomous control, unique to RIT, which could be used in a
variety of different application; including ones that could potentially be marketable. To
complete this objective our team will develop a sensor/controller network that interacts with
customized computer algorithms, capturing and processing the data using cutting-edge Artificial
Intelligence constructs. For obstacle-avoidance and path planning, multiple sensors will be
incorporated, such as: optical, radar, sonar, laser, and Global Positioning System data (GPS).
The result will be an intelligent vehicle that will have the navigating a dynamic environment.
The G-Cart team has successfully developed a completely wirelessly controlled vehicle
over the course of three months. The team was successful in altering a 91’ Geo Storm into a
vehicle capable of being controlled from several hundred yards away. Completion of this task
marks the end of phase 1 of the project. The team is actively perusing corporate sponsorship to
continue on their journey to develop an autonomous vehicle to compete in the 2005 DARPA
Grand Challenge.
1.10 Formal Statement of Work
The Executive Board has requested that the senior design team develop systems to
control the throttle, braking and steering. For the throttle and braking portion, the sub team must
deliver a “modular” control board that is capable of being adapted to one of several choices of
sport utility vehicles. The system must interact with the vehicles CAN system and I2C main
control bus for the purpose of I/O signaling. The controller will connect to a throttle by wire
system and Continental Teves brake by wire prototype. Proper resources will allow for vehicle
simulation and system testing, with out implementation in a vehicle. The controller must
maintain the desired vehicle speed within ±3 mph, except at a desired speed of zero. It is
required that the vehicle respond faster than normal human reaction delays. Through tests the
vehicles maximum performance will be determined and used to maximize vehicle response to
the controller input.
The G-Cart team will provide a cooling system inside the vehicle, but the system will be
tested at elevated temperatures due to potentially high dessert heat. The nature of the project
demands extreme controller stability as well as back up systems. To avoid serious brake
overheating, the controller must choose either braking or active throttle, not a combination of the
two. An emergency brake will provide a means of stopping the vehicle. Through the use of a
linear actuator, the vehicle can activate a “last resort” method of deceleration. Rigorous testing
both in and out of the vehicle will prove overall stability of the speed control design.
2 CONCEPT DEVELOPMENT
This chapter will cover the different concepts the team came up with. The first few
sections discuss the break down of the team. Since there are many different aspects to the
project, the team of eight engineers broke up into subgroups.
From the subgroups, ideas were generated and brought together to aid in concept
development on Bus Architecture, Vehicle Steering, and Speed Control.
2.1 Subgroup
The team was divided up to concentrate on the different aspects of the project. There were
3 different subgroups, although the Bus Architecture group had a short term focus and the team
members were assimilated into the other remaining subgroups. The subgroups are Bus
Architecture, Steering, and Speed Control.
2.1.1 Bus Architecture
The G-CART navigation computer, navigation sensors and vehicle control systems need
to have a communication bus to transfer data and commands. This subgroup was created with a
short-term goal: to design a communication network within the first few weeks of the fall
quarter. Scott Glover and Darren Rowen worked on this subgroup. They presented their
research to the RIT G-CART Club and hammered out which design would be best for the
project.
2.1.2 Steering
The steering subgroup was formed in approximately the third week to focus on the
electronics for the steering control system for the vehicle. The steering subgroup was to work
closely with the Mechanical Senior Design team. The Mechanical team was responsible for
picking out an appropriate motor, gearing and motor mounting. The steering subgroup from the
electrical senior design team was responsible for designing a motor drive and controller for
steering system. The controller for the system must be able to communicate with the guidance
computer and be able to read data from the vehicle’s CAN communication network.
2.1.3 Speed Control
The purpose of our sub team is to create a control system for braking and throttle. The
first part, braking, should be the easier of the two. Continental Teves is a company that designs
drive by wire systems for Ford and Toyota. They have asked to help implement their prototype
brake by wire system. Braking will be controlled by a pressure sensor and a pump attached to
the master brake cylinder. The vehicles CAN system will give the current master cylinder
pressure as well as deliver the desired master cylinder pressure. The velocity sub team will read
in desired velocity to our control board, make adjustment to the brake pressure and test results
for amount of error with respect to desire velocity and pressure and compensate for unexpected
results.
The second part of the task is to control a throttle sensor and throttle position motor. The
difficulties lie in the off road application. The vehicle will be subject to many road obstacles,
uneven surfaces disturbances, turning on dirt surfaces, up/downhill non-linear gradient
disturbances to the system, and trying to maintain a somewhat constant speed at the same time.
The controller will read in current speed, desired speed, and pseudo-acceleration variables via
the I2C bus. The output will control the throttle position motor via the CAN system, test results
of the position command, and adjust for unexpected results.
The controller has to be fast enough to read the I2C bus, CAN bus, make calculation,
implement results, test results, react to error, and sift the available data to acquire the needed data
for the current function.
2.2 Bus Architecture Concepts
Perhaps the most important goal of this years G-CART project was to produce a
modular design, which would allow for future additions to the project. In order for future
integrations to go smoothly, a standard communications protocol between processors,
sensors, and actuators should be established. This communications bus should allow for
simple future additions while also providing reliable and noise-free communication
between all in-car nodes. The following block diagram shows the current G-CART bus
architecture design.
Figure 2.2.1
Below is brief description of the five communication bus protocol concepts that
were seriously considered for implementation into the G-CART vehicle. Lastly, an
overall comparison of the bus architecture concepts is presented.
2.2.1 I2C
The first communications bus protocol considered was the inter-integrated circuit
(I2C) protocol, developed by Philips Semiconductor. This concept is a simple two wire
system, which can transmit addressable packets at a rate of 3.4Mbps. This protocol
allows for seven bits of addressing (128 devices), which would allow for a great amount
of future upgradeability. This protocol also has the ability to generate broadcast
messages, which are packets sent to all nodes regardless of addressing. This would allow
for emergency messages to be sent over the network, causing all nodes to act
simultaneously, rather then sending important commands one at a time to each system
node.
An I2C communications bus is currently used on existing G-CART vehicle to
handle simple communication between closely spaced processors. Therefore the G-
CART team already has a working knowledge of this communications protocol, which
should expedite the implementation of this design.
The only major concern of the I2C protocol is its lack of noise immunity and error
correction. Although no problems with the current I2C bus have been identified, it has
only been used for small distances and limited bus traffic. For the new G-CART design,
the communications bus will be run throughout the vehicle (several meters) and be
subjected to greater amounts of EMI from both the vehicle itself and from outside sources
(power lines, radio traffic, etc.).
2.2.2 USB
The Universal Serial Bus (USB) was also considered for the communications bus
protocol. USB is a robust, fast and reliable protocol that has become somewhat of an
industry standard today. This protocol would allow for new sensors or actuators to be
connected to the bus on the fly, and immediately be recognized by the system. USB also
has the benefit of having a built-in error checking and fault-handling mechanism. This
would allow for our communications bus to have greater length and handle more volatile
environments.
The major drawback to using a USB system is cost. Processors with
programmable USB ports are typically more costly than alternatives, and software
configuration is generally more difficult as well. Even though USB is an open-source
protocol it has many advanced features that would not be required by our application.
These additional features may be distracting and cause unnecessary overhead in our
vehicle.
2.2.3 PXI
PCI Extensions for Instrumentation (PXI) is a protocol that combines the high-
speed PCI bus with integrated timing and triggering lines for data capture and
comparison. Like USB, PXI devices are typically dynamically reconfigurable, which
results in a ‘plug-n-play’ system. This protocol was designed specifically for data
capture and analysis, and therefore has several unique triggering and comparing features.
It is yet uncertain whether or not these features would be used in the communications bus
of the current G-CART design, but they could potentially be of great use in future design
revisions.
PXI devices are typically very robust and high modular. Most of the PXI DAQ
board that we looked at for this design were rated for extreme temperature, vibration, and
EMI, all much outside of the range that is expected to be encountered for this project. If
a PXI bus were implemented, all processors and controller would be similar to a PCI card
(like those used in standard PCs), which would plug into specialized PXI chassis.
Again, the main drawback to this concept is cost. Both the PXI chassis and
controllers are very expensive. This concept is mainly considered because one of the
major PXI vendors, National Instruments (NI), is currently considering the RIT G-CART
team as a sponsor. If this sponsorship is obtained then the cost of the PXI equipment
would be greatly reduced and some of NI's pre-built systems could be integrated into the
G-CART vehicle.
2.2.4 IEEE 1384
The next bus protocol concept considered was the IEEE 1384 standard,
commonly referred to as firewire. This protocol is also fast becoming an industry
standard for high-speed data transfer. Firewire is typically seen in video applications
were high bits-rates are required. In addition to its speed, firewire also allows for an
extremely high number of nodes to be connected on a bus, which would allow for future
adaptability. Firewire also included noise immunity and error correction similar to USB,
which would make it durable and reliable in the G-CART application.
It is unsure whether or not the high speed of IEEE 1384 is required on a
communications bus. Currently, there are no plans to include any video-based sensors on
this network and therefore the high bit-rate may not be needed. Also, it is relatively
difficult to find cheap controllers with IEEE 1384 programmable connectors. Most
devices that include firewire interfaces are video-specific, and include features that will
not be used for a typical bus-level controller.
2.2.5 RS 485
The final communication bus protocol investigated was RS-485. This protocol is
a more robust version of the widely popular RS-232. RS-485 is a four wire design, which
broadcasts packets from a bus master and has variable message lengths. This protocol
also has error correction, although not as comprehensive as the method employed by
USB or IEEE1384. RS-485 is a popular choice for industrial communications because it
has to be ability to provide high noise immunity over long distances. This would allow
for the communications bus to be run throughout the G-CART vehicle without worrying
about maximum distance or line capacitances.
RS-485 is also considered to be a good solution because it is easily interfaced
with most microcontrollers and processors. Most devices come standard with either a
RS-232 port or a full UART. Both of these can be used to receive RS-485 signals with
little software overhead.
2.2.6 Compare and Contrast
The concepts listed above demonstrate the wide range of communication
protocols that can be implemented into the G-CART vehicle. All five of the
communication protocol concepts allow for individual and broadcast addressing,
bidirectional communication, and a high number of attachable nodes. Some concepts
have specific attributes that make them appealing, such as the specialized triggering
commands in the PXI protocol. Other concepts provide simple bus architecture with
limited overhead, such as RS-485 and I2C. Without knowing specifically what the
communications bus will be used for in future G-CART designs, it is only possible to
choose an architecture that will provide fast and reliable packet transfer.
2.3 Steering Concepts
2.3.1 DC Drive Concepts
Many different approaches were researched, although the final concepts were
restricted by the final motor selection.
2.3.1.1 PicMicrocontroller with Hall Effect Feedback
There are some alternatives to control the motor drive (bridge). One way would
be to control it using hardware approach, which is relatively more expensive and the
reliability is questionable. Using the software approach, the design will be more cost
effective. The design will be more reliable over all due to lower hardware complexity.
For the design, the PIC microcontroller was chosen due to its cost effectiveness, available
resource and documentation, and simplicity. The recommended set up for a pic
microcontroller based 3 phase BLDC motor control is shown below:
Figure 2.3.1.1.1
2.3.1.2 PicMicrocontroller with no commutation feedback
This concept is very similar to the DC drive using commutation feedback. The
main difference is that the microcontroller will energizes the coils in a fixed pattern. The
main drawback to this design is that with loading, the motor may not speed up to the
fixed pattern and a stall could result. In the start up scenario, without motor position
information, the fixed pattern would only assist in starting the motor for one third of each
excitation cycle. A primary concern is having enough start-up torque. So the loss of
start-up torque makes this design option highly unlikely.
2.3.2 Controller Concepts
The controller would have the responsibility of reading and parsing the I2C
communication from the navigation computer. In addition, the controller would have to
monitor the vehicles CAN traffic and parse out messages containing the vehicles wheel
angle. From this data, the controller would need to computer an error signal. Using the
error signal the controller would then have to compute the appropriate motor speed and
direction.
2.3.2.1 FPGA Controller
One possible approach was to use an FPGA to act as the controller. The FPGA
would have to be capable of reading and parsing both CAN network messages and I2C
network messages. The programming of the FPGA to read and parse network data from
both I2C and CAN networks could prove a difficult task. In addition, a Digital to Analog
converter would be required to provide the reference voltage for the DC Drive.
An FPGA would easily be able to implement a fixed weight FIR filter to
implement the control algorithm. In an FPGA, the ability to adapt or vary the weights
would add a tremendous amount of programming complexity.
The FPGA would have to be installed on a circuit board with auxiliary circuitry
including power supplies, external pull-up and pull-down resistors and filtering
capacitors. In many cases demonstration boards are available for purchase.
2.3.2.2 Microcontroller
The wide range of microcontrollers on the market offers an amazing range of
capability. There are microcontrollers available that are capable of reading CAN and I2C
network traffic. A microcontroller would also be capable of implementing an FIR filter.
Being the values for the weights of the filter would be stored in memory; they could be
changed or adapted without having to reprogram the device. One concern is the ability of
a microcontroller to implement an adaptive control algorithm, such as the least-mean-
square (LMS) algorithm. Implementation of an adaptive controller requires very high
speed computations in order to minimize the adaptation time. It is questionable as to
whether a microcontroller would have the processing power to implement such a control
algorithm.
The programming of a microcontroller can often require proprietary software and
hardware, adding a lot of cost to this approach. Most microcontrollers can be
programmed via writing C or assembly code and then downloading it to the
microcontroller.
The microcontroller would have to be installed on a circuit board with auxiliary
circuitry including power supplies, external pull-up and pull-down resistors and filtering
capacitors. In many cases demonstration boards are available for purchase. In addition, a
Digital to Analog converter would be required to provide the reference voltage for the
DC Drive.
The validation of such a controller would be very involved. External
communications monitoring hardware would most likely be necessary. Troubleshooting
could prove very difficult for a microcontroller based controller.
2.3.2.3 National Instruments Real Time Controller
National Instruments offers a Real Time Control module for the PXI platform.
This module uses the LabVIEW Real-time operating system. There are several different
models available, which all have different features. There is a module available that has
an extended temperature operating range and a tremendous amount of processing power.
The extended operating range would work well for our application. These modules are
designed for an industrial environment and would offer many advantages in terms of
reliability.
The programming of the Real Time Control Module is done using a standard
Ethernet link. After the code was written in LabVIEW on any windows or Macintosh
computer it can be written via the Ethernet link to the control module.
The control module would have to be installed in a PXI chassis. These chassis are
designed for modular components. Other modules, such as a CAN communication card
could simply be snapped into place. No external wiring or circuitry would be required.
For reading I2C and CAN traffic there are PXI modules and drivers readily
available. The CAN module offered by National Instruments comes with software
capable of parsing data.
The validation and testing of the controller and steering system would be made
much easier by using a National Instruments Controller. National Instruments originally
started out in the electronics testing business and have since expanded. The built in
troubleshooting and test capabilities could prove invaluable.
Although National Instruments equipment is typically expensive, we may be able
to obtain their sponsorship. The company’s interest in partnering with RIT will most
likely allow for the donation of the majority of the equipment required.
2.4 Velocity Control Concepts
The main focus of the velocity control subgroup is to develop a method to reliably
and quickly alter the G-CART vehicle’s speed. This involves interpreting set points
handed to the communications bus by a higher level navigation computer, and making
corresponding changes to a the vehicle’s throttle and braking systems. For the proposed
G-CART vehicle, both throttling and braking systems are done ‘by-wire’ and involve to
mechanical actuation. Below is a description of controller concepts investigated by the
velocity control subgroup. An analysis of the feasibility of each control method is shown
in section 3.3.
2.4.1 Controller Concept
After much research, it became obvious that a PID controller was necessary as the
controller for the throttle/braking portion of the autonomous vehicle. It was decided that
our system should work similar to how a cruise control system would work. The only
difference being that overshoot would not be a problem as long as the response time was
better and the setting time was reasonable.
To understand why a PID controller was necessary, it is important to understand
what a PID controller is. PID stands for proportional-integral-derivative, each of which
makes up the three parts of the controller. For a PID controller, there is usually a desired
set point that the controller must meet. It is based on error between the desired set point
and the output of the system. The proportional part of the controller is just the error
multiplied by the gain Kp. The integral portion is the integral of the error multiplied by
the gain Ki. Finally, the derivative portion of the controller is the rate of change of error
multiplied by the gain Kd. Kp, Ki, Kd are gains that can be changed in order to meet the
optimal response.
The most basic control system would be the proportional controller. Basically,
what the proportional controller would do is adjust the throttle proportional to the error.
Therefore, the greater the error became the greater the throttle. The problem with that
type of system would be that the closer the car was to the desired velocity, the slower the
car would accelerate and when a fast response time is desired, this type of system is not
reasonable. Also, if the car was traveling up a hill that was steep enough, it may not
accelerate at all.
The next type of controller is the proportional-integral controller. This type of
controller does everything that the basic proportional controller can do and more. The
integral of the speed is distance. Therefore, since the integral of the error is taken, the
integral portion of the controller gives the difference between the distance that the car
should have traveled if it were to have traveled at the desired speed and the distance that
it has actually traveled. This corrects problems that may arise when traveling up hills and
also helps to settle the car into the correct speed and stay there. The integral portion of
the controller removes the need for a tilt sensor to provide another signal to the controller.
It also makes the programming for the controller less complicated.
The final type of controller is the proportional-integral-derivative controller. The
derivative of speed is acceleration. Therefore, it can be seen that the derivative portion of
the controller will affect the acceleration of the vehicle. The derivative portion of the
controller will help the car to respond quickly to changes. For instance, if the car begins
to slow down due to a hill, the controller will see the downward acceleration before the
speed of the vehicle changes significantly, and will therefore respond by increasing the
throttle.
While a proportional or a proportional-integral controller would probably be able
to control the car, a PID controller has all the qualities that the desired vehicle should
have in order to compete. Not only do we want the car to be able to get to a desired
speed, but we want it to reach that speed as fast as possible to ensure that the car is as
close to where it should be as possible.
2.4.2 FPGA Concept
FPGAs are digital ICs (integrated circuits) that contain configurable blocks of
logic along with configurable interconnects between blocks. Design engineers can
program this kind of device to perform a remarkable variety of tasks. Field
Programmable means that the device has not been hardwired by the manufacturer, which
creates the flexibility. Some FPGAs can only be programmed once, while others can be
reprogrammed many times over. This would be similar to CD-R verse CD-RW for data
storage on a disk. One-time programmable (OTP) is the implicit term used in reference
to FPGAs.
FPGAs create a middle ground between ASICs (application–specific integrated
circuits) and PLDs (programmable logic device). An FPGA can be used to implement
large and complex functions that previously had to be realized by ASICs. The cost of
FPGA design is much lower and easier to manipulate for design changes. This means
that small engineering groups can realize hardware and software on a test platform
without major fabrication costs. Some FPGAs contain millions of gates, and can offer
high speed Input/Output interfaces, digital signal processing (DSP), and reconfigurable
computing (RC). RCs are used as a “hardware accelerator” for software algorithms.
FPGA devices are capable of being programmed while remaining resident in a higher-
level system; this is referred as being in-system programmable (ISP). Altera’s Nois II
board was donated to our team as a possible means for a speed control solution.
Figure 2.4.2.1
Nios II Processor System Basics:
The Nios II processor is a general-purpose RISC processor core, providing:
■ Full 32-bit instruction set, data path, and address space
■ 32 general-purpose registers
■ 32 external interrupt sources
■ Single-instruction 32 × 32 multiply and divide producing a 32-bit result
■ Dedicated instructions for computing 64-bit and 128-bit products of multiplication
■ Single-instruction barrel shifter
■ Access to a variety of on-chip peripherals, and interfaces to off-chip memories and peripherals
■ Hardware-assisted debug module enabling processor start, stop, step and trace under integrated
development environment (IDE) control
■ Software development environment based on the GNU C/C++ tool chain and Eclipse IDE
■ Instruction set architecture (ISA) compatible across all Nios II processor systems
■ Performance beyond 150 DMIPS
One of the notable features of the Nios II processor is termed Configurable
Soft-Core Processor. The CPU is in a “soft” design form, contrary to fixed
microcontrollers. Essentially, it is a blank FPGA. This allows the user to configure the
processor and peripherals to meet their needs and then program the system into an Altera
FPGA. Altera offers some ready made Nois II systems design, so that the user does not
have to create a new processor configuration for every design. A Flexible peripheral set
and address maps allow for an exact peripheral set intended for the target application.
Furthermore, Altera’s SOPC Builder design tool fully automates the process of
configuring process features and generating a hardware design that can be programmed
into an FPGA. After the system generation and programming the board, the software can
be debugged through on board execution.
One of the Key ideas discussed in the Quartus II handbook is the
significance of timing in relation to system reliability. A synchronous design implements
a clock signal trigger for all events. All of the registers’ timing requirements must be
must as well. Without such a design the system may be dependent on propagation delays
and create possible glitches. Typically, the data inputs of registers are sampled and
transferred to output on every active rising edge. The outputs of combinational logic
feeding the data inputs of registers will then change values. The internal circuitry of the
registers isolates data output from inputs; therefore instability in combinational logic does
not affect the operation of the design as long as two things are considered. First, the data
input must be stable for the setup time of the register (before an active clock edge).
Second, the data must remain stable for the hold time of the register (after an active clock
edge). If the setup (tSU) or hold time (tH) is not met, output can be set to an intermediate
level between high-low values. In this unstable state, small disturbances, like noise in the
power rails, can cause registers to assume unpredictable valid states. When controlling
the speed of an autonomous vehicle, unpredictable valid states are not an option. The
throttle could be held in an “on” position, the brakes could malfunction or the brakes and
throttle could be active at the same time. System stability is a requirement for a speed
FPGA controller.
2.4.3 8051 Concept
One of the other potential controller options is a Rigel Development
Board, provided through the Electrical Engineering Department at RIT. This board uses
a C515C processor, which is a subset of the Intel 8051. Because this board uses an
industry standard processor, there is a considerable amount of pre-existing libraries and
functions that could potentially simplify controller design.
This control board is used in several RIT courses, and members of the
velocity control subgroup already have experience using this software, which should
reduce the difficulty in implementing this design. In addition, this controller design also
includes 48 digital I/O lines and eight analog inputs. Although the current design should
not require more than four digital I/O lines, this adaptability could provide useful for
testing and data logging purposes. This control board also has
An on-board Full-CAN controller, which would allow for decoding of message from the
automotive CAN bus without extensive software programming.
An on-board DUART, which would allow for simple connection to either an RS-232/485
or I2C communication bus
10-bit A/D converter for analog actuator control
Multiple priority interrupts, which could be manipulated by the controller bus masters in
order to change message priorities
Although the current Rigel boards are available free of charge for this project,
they are somewhat obsolete. They operate at 12MHz, with most operations taking 12
machine cycles. This speed is notably slower than the FGPA alternative, and may limit
the effectiveness of this solution. There are newer control boards produced by Rigel that
operate much faster, and are completely backwards compatible. Therefore, if a controller
was designed on this board and it was determined during the testing stage that the
response was too slow, then faster boards could be purchased (~$200) with no software
changes required.
2.5 Emergency Brake Concepts
The E-brake concepts that were considered are divided into two parts, the actual
method of actuating the brake and the type of actuator to be used. The combinations of
these two parts also play a big role on each other, which will be discussed in the next
sections.
2.5.1 Actuating method
The first method is to pull the e-brake from behind as seen in Figure 2.5.1.1 with
the other end of the actuator mounted to the floor panel. This design allows for the
mechanical advantage of the lever to be used but the disadvantage of this is that the
stroke of the cylinder will have to be longer which causes a longer response time for
actuators with slow travel speeds. After testing and taking the worst case of five different
vehicles I came up with a stroke of approximately 4 inches and a load of roughly 130 lbs
for this method of actuation. The biggest benefit of this design is that it is the simplest
and most versatile of the two and probably the most reasonable of the two since the exact
application is unknown, so the system that is designed will have to be easily modified to
meet the vehicle needs.
Figure 2.5.1.1
The second method is to remove the brake lever all together and pull directly on
the cable without any mechanical advantage which can be seen in figure 2.5.1.2. This
design will have a very short stroke compared to the first concept which will lead to a
quick response time which is a very large advantage. The downfall to this is that since we
are not using the mechanical advantage this short stroke is going to have a very large load
which makes the actuators more expensive and less compact. The stroke requirement for
this application is roughly 1.5 inches and a load of roughly 450 lb.
Figure 2.5.1.2
2.5.2 Type of Actuator
There are three types of actuators that were considered air, electrical, and
hydraulic systems. They all have there advantages and disadvantages. And depending on
the method of actuating the brake they also have there advantages and disadvantages.
The air system has very fast stroke rates with high loads which is essential when
time is one of the biggest factors. Air cylinders are also a very simple mechanical device
and therefore also very robust but for this application it is not a very crucial attribute.
Since we will probably only actually use the actuator a few times if any. The downfall of
the air system is the complexity of the overall system and the time it will take to install
the system. When looking at the price of the system it is not far behind an electric system
since there are many more components in the air system, like air compressor, tank,
electric valves, etc.
The hydraulic actuator has many of the same attributes the air actuator has. The
biggest problem with the hydraulic actuator is we don’t have the expertise to install a
system like this and since it is allot like the air the air actuator will be chosen over the
hydraulic actuator.
The electrical actuators are a little more than the air system if you want good
response times but if time could be sacrificed the electrical actuators are much cheaper
than the air systems. Some very good attributes to the electrical system are there
versatility and there ease of installation, which with the time constraint that we are under
is very good since we need time to install theses systems and debug them.
3 FEASIBILITY
Feasibility studies were preformed on many of the design concepts to guide our
team on the direction to proceed. Pugh’s Method and the weighted method were both
used as references while performing the feasibility assessment.
3.1 Bus Architecture Feasibility
The five bus architectures described were analyzed through the use of a weighted
method, which took into account the nine indicators shown on the table below. Each
characteristic is assigned a weight between zero and ten, which is multiplied by the score
given to a particular concept. The sum of all the individual indicators gives the final
feasibility of each concept.
Defining Bus Characteristics
Rel
ativ
e
Wei
ght
I2C
USB PX
I
IEE
E13
84
RS
485
Bus Speed 5.0 5 8 7 10 4
Number of Nodes on the Bus 3.0 5 5 3 8 7
Error Handling 5.0 0 8 4 8 4
Noise Immunity 3.0 3 9 8 7 8
Robustness of Design 5.0 4 7 10 7 8
Simplicity of Design 5.0 10 5 2 2 8
Cost of Components 6.0 9 6 1 4 8
Difficulty of Integration 8.0 9 5 7 2 8
Team Familiarity 8.0 10 2 2 2 5
Score (sum of weighted values) 325.0 274.0 226.0 236.0 317.0
Normalized Score 1.000 0.843 0.695 0.726 0.975
Table 3.1.1
This weighted method shows that the I2C method of communication is the most
effective for our application. This is mainly due to the low cost, ease of integration with
existing components, and previous G-CART team experience with the protocol. The
IEEE1384 and PXI concepts were identified to be poor alternatives due to high
complexity and difficulty of integration.
3.2 Steering Feasibility
3.2.1 DC Drive Feasibility
The DC Drive must be capable of driving the Brushless DC motor chosen by the
mechanical engineering steering team. The BLDC motor demands that the drive be able
to supply 36V and peak current of 38.7A. The motor is 3 phase, so the drive must be
capable of driving a 3 phase motor. The Brushless DC motor does not operate directly off
a DC voltage source; however, it has a rotor with permanent magnets, a stator with
windings and commutation that is performed electronically. Typically three Hall sensors
are used to detect the rotor position and commutation is performed based on Hall sensor
inputs.
The team members have very limited knowledge with motor drives and therefore
a more feasible solution was to purchase or request a high power commercial BLDC 3
phase motor drive. The two drives that the team considered were MKS 4351 and aeroflex
ACT5101 both of which is capable of operating at 50A/500V. Some research went into a
lower voltage model, but none were found available.
Constructing a BLDC drive from scratch would likely be a more cost effective
solution, but as mentioned earlier, with limited design time and capability of the motor
drive, it is not feasible to design one from scratch.
3.2.2 Controller Feasibility
The controller concepts described were analyzed through the use of a weighted
method, which took into account the nine indicators shown on the table below. Each
characteristic is assigned a weight between zero and ten, which is multiplied by the score
given to a particular concept. The sum of all the individual indicators gives the final
feasibility of each concept.
Defining Controller
Characteristics Rel
ativ
e W
eigh
ts
FPG
A C
ontr
olle
r
Mic
roco
ntro
ller
NI R
eal T
ime
Con
trol
ler
Processing Power 6 8 7 9
Testability 8 3 3 10
Troubleshooting Capability 8 4 4 10
Programming Difficulty 9 3 5 10
Robustness of Design 6 7 7 8
Simplicity of Design 3 3 4 9
Cost of Components 4 10 7 4
Difficulty of Integration 5 3 4 9
Team Familiarity 5 3 5 7
Score (sum of weighted
values) 252 270 475
Normalized Score 0.53 0.57 1.00
Table 3.2.2.1
This weighted method shows that the National Instruments Real-Time Controller
is the best choice for our application. This is mainly due to the Testability, trouble
shooting capability and programming ease. The availability of drivers and modular
communication cards makes this system very easy to integrate. Being we have not yet
heard a final word on the sponsorship of these components the Cost was rated at 4/10. If
the sponsorship comes through, this would increase this score and make this design even
more feasible. It should be noted that the same type of control algorithm could be
implemented in any of these designs. So although one concept is more feasible, all of the
concepts looked at are certainly capable of fulfilling the needs of this project.
3.3 Speed Control Feasibility
3.3.1 Controller Feasibility
The greatest advantage of a PID controller is its robustness. The three different
error control operations of the controller help to rapidly correct any deviation between the
desired speed and the actual speed. The controller’s ability to quickly correct these
deviations is necessary for proper function of the autonomous vehicle. However, a major
draw back to the PID controller is situations that can disrupt the effectiveness of the
derivative portion of the controller.
There are two situations that occur that causes problems with the derivative
portion of the controller. The first is any sharp change in the desired velocity with
respect to the current velocity. This sharp change would result in an unreasonable size
control input to the vehicle. This may not be a problem for our vehicle due to the fact
that we do not anticipate any dramatically sharp changes in the velocity. The stability
system on board should stop any slippage from causing a spike in the velocity measured.
Also, it is expected that the any changes in velocity would be received from the on board
computer in smaller increments to insure smooth transitions in speed. The second
situation that may be more of a problem is noise. Noise may cause undesirable spikes in
sensor information that would be fed into the controller. Like in the previous situation,
the sharp change would result in an unreasonable size control input to the vehicle. This
could possibly pose a problem for our system.
At this point, even with the problems, a PID controller may still be the best
option. Current cruise control systems use PID controllers as a means to adjust velocity
without many problems. For this reason, a PID controller remains the number one
option. A PI controller may work as an alternative, but it would not be as robust and as
quick to respond to changes, but would accomplish the task.
3.3.2 Velocity Control Feasibility
For our purpose, an FPGA controller seems ideal with respect to the design task at
hand. FPGAs are adaptable, more than fast enough, free of cost to us, and expandable to
meet unforeseen needs. The problems do not lie in the technology, but are inherent in our
team. Our team does not have a license for the software that was provided with the x-
caliber FPGA. Another problem is the time limitations set forth by the Fall/Winter
blocks. None of our team members are familiar with VHDL, and we would not want our
project to fail based on our lack of VHDL knowledge. There are several other options
that can be used to implement a solution, and will be discussed further.
In order to give our team time to acquire a software license and to become
familiar with the VHDL language, the 8051 development board will be implemented as a
primary solution. This development board has more than enough speed to parse the
income I2C and CAN data-streams. At the time this primary software development is
completed, it will be determined whether or not we can use the FPGA to perform the
actual PID control. If the FPGAs are deemed feasible, we will use it in conjunction with
the code already developed on the 8051 board. If the FPGAs are not feasible we will
perform the PID control on the 8051 development board itself.
3.3.3 Emergency Brake Feasibility
The feasibility assessment for emergency brake was assessed using a radar chart
with the key attributes being cost, versatility, durability, reaction time and ease of
installation. In this method the polygon that has the greatest area is the most feasible
system. From this radar chart below it can be seen that the electrical system is the most
suitable for this application. With the air system not far off, but as discussed earlier when
considering versatility, and being able to easily bench test the system the electrical is the
best choice.
01234
low cost
high versitility
high duribilityshort reaction time
easy overall installation Air
ElectricalHydraulic
Figure 3.3.3.1
3.4 Feasibility Conclusion
For the bus architecture, the team selected the I2C protocol. As shown in our
feasibility section, this design proved to be the best fit for the G-CART vehicle. It was
primarily because of the simplicity of this design, as well as our familiarity and
experience with this protocol that it was found to be better than the other alternatives.
Although this protocol does not offer noise immunity or contain internal error checking,
it is believed that this drawback can be overcome by the use of shielded cable and
additional EMI reducing devices. It is not believed that there will be high amounts of
EMI generated by any component of the G-CART vehicle, nor will there be high levels
of interference present in the desert terrain of the course.
The steering subgroup has decided to implement their control algorithm on a
National Instruments Real-Time controller module. This concept was selected because of
its high testability, ease of programming, and robust design. The mechanical G-CART
senior design team (5106) was responsible for selecting the motor to drive this
subsystem. Because no such decision has yet been made, the steering team also chose a
highly adaptable power regulator for this system. The MKS 4351 was selected mainly
because it provides enough power to drive even the most demanding of the concepts
proposed by the 5106 team.
The speed control subgroup has decided to use a hardware implemented PID
control method. Ideally, this control will be performed on the National Instruments Real-
Time controller module. However, until sponsorship with NI is obtained, the team is
current designing an Intel 8051 based solution. A detailed description of the 8051
development board is given in the above speed control concepts subsection. The speed
control group also has decided to pursue the electrically actuated emergency braking
system. Compared to the other braking alternatives, the electrical method was much
simpler to install and provides for simpler bench testing.
4 OBJECTIVES & SPECIFICATIONS
A set a guidelines, design objectives, and performance specifications was
established to assist the team in properly assessing on how successful the outcome of the
project is. The following section of the chapters will go through the different objectives,
specifications, and guidelines that the team agreed upon.
4.1 Design Objectives
There are a number of design objectives that required the attention of the team.
These objectives have to be specified in order for the team to have a list of goals and aims
to achieve. These objectives are listed below:
1) The first objective of the design project is to design a system that will be able to
receive desire velocities and steering angles from the G-Cart club’s bus master
through I2C bus protocol.
2) The second objective is to design the system such that it will also be able to receive
sensory data from the car’s sensors and encoders through CAN bus protocol.
3) The third objective is to design the system in a way such that the system, given the
input above, will be able to drive the steering motor to turn the steering wheel to the
desired angle positions.
4) The system, given the input above, should be able to control the car’s acceleration
and braking system to achieve the velocities requested by the navigation computer.
5) The systems should be designed in modules for the purpose of modular testability and
for the ease of system expansion.
6) The final objective is to ensure that the systems operate reliably and swiftly.
4.2 Performance Specifications
The DARPA Grand Challenge is an autonomous vehicle competition. The
contest requires contestants to successfully create an autonomous vehicle capable of
traversing a timed 175 mile off road desert race. Additionally, vehicles must be able to
stay within given GPS boundary points while avoiding various natural and man-made
obstacles without causing any damage to the course. While the overall task may seem
daunting, our concern is on a much smaller scale. The scope of this project required the
design considerations of three separate entities. These three entities were the steering
control system, the speed control system, and the communication bus that connects the
Navigation computer to the various subsystems.
4.2.1 Steering control performance specifications
4.2.1.1 The response of the steering system, at a minimum, shall respond as quickly as a human operator.
4.2.1.2 The steering system shall be capable of controlling a wide selection of three phase DC brushless motors.
4.2.1.3 The system shall be flexible enough to be able to easily be installed in comparable SUVs.
4.2.1.4 The steering system shall be capable of parsing I2C and CAN network messages.
4.2.1.5 The steering system shall minimize the error between the desired wheel angle and the actual wheel angle.
4.2.2 Speed control performance specifications
4.2.2.1 The Speed control system shall be capable of parsing I2C and CAN network messages.
4.2.2.2 The speed control system shall minimize the error between the actual speed and desired speed.
4.2.2.3 The system shall effectively stop the vehicle in response to a zero velocity command.
4.2.2.4 The system shall be capable to use the emergency brake to stop the vehicle in the case of power loss.
4.2.3 Communication bus performance specifications
4.2.3.1 The communication bus shall robustly transmit data between the navigation system and subsystem.
4.2.3.2 The communication bus shall support the bandwidth required for the network traffic.
It is inevitable that as the project continues, the team will face numerous obstacles
and problems. However due to time constraints, not every issue will be addressed. By
having a list of performance specifications, it will aid the team in prioritizing what is
crucial. This will help manage time more wisely into what problems must be fixed and
which obstacles the team can overlook.
4.3 Safety Issues
To ensure the safety of team members and non members, a set of safety
precautions was established. The project consisted of both electrical and mechanical
system and therefore, the safety precautions cover both aspects of the system.
The design and experiment will be conducted in a closed area. Extreme
precaution must be given when dealing with the steering electrical system, because high
current is drawn to the steering motor. The system should be unplugged, before
mechanical modification is made to prevent injuries. Mechanical modification should
only be done when 2 more members are present in the experimental area.
5 DESIGN ANALYSIS & SIMULATION
5.1 Steering Design and Simulation
The overall system block diagram can be seen in Figure 1.
Figure 5.1.1: Steering System Layout
5.1.1 Motor Power Analysis
The maximum power requirement based on the motor selected by the mechanical senior
design team was 1393Watts. This is an extremely high power requirement. However, we
should never exceed this value. For normal operation, we would only require less than
80% of maximum Power. Additionally, this motor will require a 36V supply.
5.1.2 Motor Analysis
There are many types of motors that can be used for steering, such as:
2 Phase DC Motor (with Brush/ Brushless)
3 phase DC motor (with Brush/ Brushless)
Multi–phase DC motor (with Brush/ Brushless), which is similar to a stepper
motor.
Stepper motor
The brush motor concept does not rely on controlled commutation to run, because
the brushes provide a mechanical commutation. However, this motor does not have as
high reliability as a brushless DC motor. Brush life may limit the lifespan of a Brushless
DC motor. The Brushless DC motor does not have the same lifespan concerns due to the
absence of mechanical contact inside the motor. An extended reliability is thus obtained.
A three phase Brushless DC motor would provide a more even torque distribution
than a two phase DC motor. In addition, an increased accuracy would be achieved.
Parameter Symbol Units Value
Design Voltage V volts 36
Continuous Stall Current1 IC amperes 12.3
Peak Current2 IP amperes 38.7
Voltage Constant +/- 10% KE V/kRPM 16.3
Torque Constant +/- 10% KT oz-in/amp 22
Resistance +/- 10% RM Ohms 0.6
Inductance LM mH 1
Table 5.1.2.1: Motor Parameters
5.1.3 Motor Commutation Analysis
Unlike a brushed DC motor, the commutation of a BLDC motor is controlled
electronically. To rotate the BLDC motor, the stator windings should be energized in a
sequence. It is important to know the rotor position in order to understand which winding
will be energized following the energizing sequence. Rotor position is sensed using
embedded Hall Effect sensors. Whenever the rotor magnetic poles pass near the Hall
sensors, they give a high or low signal, indicating the N or S pole is passing near the
sensors. Based on the combination of these three Hall sensor signals, the exact sequence
of commutation can be determined.
Based on the physical position of the Hall sensors, there are two versions of
output. The Hall sensors may be at 60° or 120° phase shift to each other. The motor
used for this design supplies Hall sensors at 120° phase shift to each other.
Each commutation sequence has one of the windings energized to positive power
(current enters into the winding), the second winding is negative (current exits the
winding) and the third is in a non-energized condition. Torque is produced because of
the interaction between the magnetic field generated by the stator coils and the permanent
magnets. Ideally, the peak torque occurs when these two fields are at 90° to each other
and falls off as the fields move together. In order to keep the motor running, the
magnetic field produced by the windings should shift position, as the rotor moves to
catch up with the stator field.
In the figure below, we can see Hall Sensor Output over two electrical cycles.
The Back EMF will not be measured or used in the Hall Sensor based design.
Alternative motor commutation would be via an encoder. Encoders can resolve
the position of the motor usually much more accurately than a Hall Sensor system.
However, this added resolution is not that useful to the DC drive. The DC drive only
requires a rough position in order to excite the coils in the proper sequence. Encoders are
typically much more expensive when compared to the simplicity of the Hall Sensor
commutation. Most motors do not come standard with an encoder present. Sometimes
encoders can be purchased as an upgrade.
5.1.4 DC Driver Analysis and Design
There are many different type of drivers out there can be used to control 2, 3 or
multi-phase DC brushless motor. However, the concept of driving is similar. Some
drivers use MOS and some use BJT, the below figure is one example.
Figure 5.1.4.1
Additionally, each driver is integrated differently; some have multiple inputs,
outputs, and Voltage supply requirements. Fore example, driver MSK4351 has 2 different
voltage supply requirements.
Figure 5.1.4.2
5.1.5 Microcontroller Selection
In making the microcontroller selection, the team member first considered the
constraints of the project. Design time and Cost were among the highest priorities that
must be taken into account. Material that may affect the design time includes
documentations, available technical support, and available information and application
notes. Due to the lack of funding of the project, the cost of the microcontroller also
played a major factor in the selection process. The second constraint that was considered
was availability. The product should be in production and available upon order. The
product should arrive at the design facility within a week of ordering. Finally the
performance of the microcontroller was important, but as long as the microcontroller
performance meet the need of the project it is acceptable.
The research carried out by the team members revealed that the three major
manufacturers of motor control microcontrollers were Microchip, Atmel, and Motorola.
Upon further investigation, research finding that microchip’s PIC microcontroller was the
best choice for the project. First of all, all the microcontroller cost about the same, less
than 5 dollars a unit, but PIC’s documentation, information, and application notes were
more readily available than the competitors, which will reduce much of the team’s project
design time. The competitor’s units that were considered were Atmel’s microcontroller
that was using 8051 architecture and Motorola MC68HC based MCUs. To be more
specific, the PIC unit chosen was the PIC18 family product.
Some spec of PIC18 as listed on Microchip’s website as
- Power Control PWM (PCPWM)
- Up to 8 output channels
- Up to 14-bit PWM resolution
- Center-aligned or edge-aligned operation
- Hardware shutdown by Fault pins, etc.
- Quadrature Encoder Interface (QEI)
- QEA, QEB and Index interface
- High and low resolution position measurement
- Velocity Measurement mode using Timer5
- Interrupt on detection of direction change
- Input Capture (IC)
- Pulse width measurement
- Different modes to capture timer on edge
- Capture on every input pin edge
- Interrupt on every capture event
- High-Speed Analog-to-Digital Converter (ADC)
- Two sample and hold circuits
- Single/Multichannel selection
- Simultaneous and Sequential Conversion mode
- 4-word FIFO with flexible interrupts
5.1.6 Microcontroller Programming Synthesis
The steering system utilizes a closed loop control system due to various benefits
over open loop control systems. The use of closed loop control system utilizes feedback
that will increase accuracy and stability of the control system.
The programming synthesis procedure will follow the recommendations from
various application notes from Microchip. Some application notes that can be referred to
during the programming synthesis procedures are:
AN899 for dc brushless motor control using PIC18FXX3.
AN885 BLDC motor fundamentals.
The basic program flow chart can be found in appendix 9.1.
The PIC will be programmed using MPLAB software available in the robotics lab.
5.1.7 Controller Design
The controller designed is an Adaptive Model Controller. The adaptive model in
Figure two is an FIR Filter of length L. Instead of having a fixed linear compensator
where the weight values are not functions of the input-signal characteristics, the adaptive
process automatically adjusts the weights so that for the given input-signal statistics, the
model provides a best minimum-mean-square error fit to a sampled version of the
combination of the zero-order hold and the plant. The LMS (Least-Mean-Square)
algorithm will be employed to compute the weights in the vector W(k).
Figure 1: Block Diagram of the Adaptive Model Control System
Equation 1: Adaptive Model
Equation 2: Forward Time Calculation
Equation 3: LMS Error Signal
Equation 4: LMS Algorithm
Note that μ in Equation 4 is the gain constant that regulates the speed and stability
of adaptation. The larger μ becomes, the faster the adaptation occurs. However, this
increase in adaptation time is counteracted by a decrease in system stability. In practice,
finding μ is an iterative process. Typically a very small number is used as an initial guess
and then system testing should be preformed with different values of μ to determine
stability tolerance.
The forward time calculation is designed to derive x(k) from r(k) such that r(k)
and y(k) are equal. If r(k) and y(k) are equal, then the plant output, g(k), will be close to
r(k), obtaining the control objective.
5.1.8 Controller Communication
5.1.8.1 CAN Traffic Decoding
The National Instruments Real-time controller supports pre-written
drivers for CAN communication. The only hardware that required would be a
PXI CAN communications module. The programming complexity of reading
the desired messages and parsing them would be minimal.
5.1.8.2 I2C Message Decoding
The Decoding of the I2C network traffic would be more difficult
because the message format is a G-CART proprietary design. Therefore some
decoding software would have to be written in order to obtain the correct
command signals.
5.2 Speed Control Design and Simulation
The design process is still happening, as is the case with most projects. The more
that is learned, the better the design gets. The team is currently investigating several
methods to implement our design and trying to determine feasibility. The
communication bus between the main navigation computer and bus line controllers will
be I2C. The vehicle will have a throttle by wire system and a brake by wire system will
be installed by Continental Teves. The vehicle CAN bus will provide a means of data
acquisition, as well as a means of sending CAN control messages to the master brake
cylinder and throttle position motor. The CAN system provides data updates every seven
milliseconds, which is more than fast enough for our expected needs. The car can not
react faster than the CAN system updates, thus we should have enough time to give
commands, test results and correct for over/under compensations of the PID controller.
Figure 5.3a shows a comprehensive system overview.
FIGURE 5.3a – Expected data flow for speed control
An important part of this design is to ensure timing control, since multiple
readings are coming from the same CAN system. Each reading from the CAN bus could
happen before or after a speed adjustment by the controller. The buffers are wiped
(loaded with the most negative number of the range) before every seven millisecond
CAN update, so the controller can determine relevant data. If the system is in a state of
no change the buffers will contain data out of the usable range. When there is a
significant error, it will be added to the desired change in speed. The greater the required
change in speed, the faster the vehicle will try to attain the desired speed. The PID will
create its own acceleration rates. The acceleration adjustment will become relevant in
such instances as steep inclines and decline. The goal is to achieve the desired speed as
fast as possible and to maintain constant speed when it is requested.
To get a feel for a vehicle response, our team manipulated the Simulink automatic
transmission model. The model allows for several modifications that can be help model
the performance specs of the Toyota Sequoia. Since there is no vehicle available to us,
we need to understand the throttle and braking reactions to given input. Figure 5.3b
shows the model diagram. The automatic transmission model yields results for passing
maneuvers, gradual acceleration, hard braking, and coasting functions. Output is given in
terms of engine RPMs and vehicle speed.
FIGURE 5.3b – automatic transmission simulation model
FIGURE 5.3c – vehicle input for coasting and hard braking.
FIGURE 5.3d: vehicle output for coasting and hard braking - y-axis is speed in mph
(yellow) and 25% throttle (purple). The x-axis is time in units of seconds. The vehicle throttle is active from 0-5
seconds followed by a coasting period from 5-25 seconds. The brakes were applied at the 25 second mark as
indicated by the increased negative slope.
The goal is to acquire the real vehicle response through data logging. There are
data logging tools available that will tap into the CAN system and record the response to
test drive conditions. Continental Teves has also stated that they are willing to share the
vast test data they have. The data in conjunction with a CAN simulator should allow
sufficient vehicle modeling in the lab. The hope is to have a functioning system by the
time the G-Cart team receives an SUV. The controller will only have to be tweaked for
specific vehicle performance.
The initial steps are to get the CAN and I2C simulator to properly talk to the 8051
board. The focus will then turn to modeling vehicle performance. At this point the speed
of the 8051 board will be assessed for proper response time and overall performance.
The 8051 board allows for easy processor upgrades, if the performance is not
satisfactory. Parallel to the 8051 design will be an FPGA design attempt. The lack of
experience with FPGA design leads to its secondary position in the design schemes. If
our investigation of FPGA design proves feasible, within the time constraints, than it may
become the primary design. A third possibility lies in an 8051/FPGA combination
design. For now the speed control team will stick with the 8051 implementation. The
message structure for the I2C bus was determined by the Executive Board for compliance
with there systems, and CAN protocol is standardized. Continental Teves will provide
vehicle specific addressing for their systems, based on the vehicle acquired by the G-Cart
team (and Al Simone).
Another consideration is with respect to PID verses a PI controller. The 8051
board uses a derivative of Assembly Language. Current efforts are focused on the PI
control program development. Upon completion, need for PID system stability will be
determined. It may be difficult to conclude with out the specific vehicle at our disposal.
The intent is to prove viability without a vehicle.
6 FUTURE PLANS
6.1 Schedule
The following chart illustrates the proposed timeline for this coming winter. The
schedule shows projected start and finish dates for each activity. This schedule assumes
that an exact vehicle will be obtained by the start of the next quarter.
Figure 6.1.1
6.2 Budget
Currently RIT has given a grant of $5,000 for the G-Cart team which $2,000 has
been allocated to both the navigations and controls senior design teams with the rest
being used to promote extra funding. Therefore, we are trying to get many of our part
donated, so that this budget can be met.
7 CONCLUSION
The senior design team focused the six facets of design: Recognizing and
Quantifying Needs, Concept Development, Feasibility Assessment, Design Objectives
and Performance Specifications, Synthesis and Analysis, and final Preliminary Design.
The goal of this senior design team was to provide engineering assistance to the
RIT G-CART club in the development of vehicle steering and speed control. Our
engineering knowledge as 5th year electrical and mechanical students proved to be very
helpful to the RIT G-CART club which has many younger members. Our senior design
team was able to bring a more formal design process to the table and help to organize the
different facets of the design process for the greater G-CART project.
Extensive research and concept development efforts were undertaken. Our design
process was forced to be very flexible as many factors were influencing our design
choices. The RIT G-CART club is still in the process of obtaining a vehicle for the race.
They have narrowed the choices to a few select models. This long selection process has
increased the need for design flexibility and modularity.
Going into the winter quarter, there is a tremendous amount of work left to be
done. This quarter has proven to be a project that required more expertise in team work
and project management than in actual hardware design. There are many challenges
associated with working in such a large team effort.
8 REFERENCES
[ 1 ] Microchip Technology Inc, Application note AN899 – “Brushless DC Motor Control
Using PIC 18FXX31” (DS00899)
[ 2 ] Microchip Technology Inc, Application note AN857-“Brushless DC Motor Control Made
Easy” (DS00857)
[ 3 ] Microchip Technology Inc, Application note AN885-“Brushless DC Motor
Fundamentals”(DS00885)
[ 4 ] Brushless DC motor drive for steer-by-wire and electric power steering applications.
Rodriguez, F.; Uy, E.; Emadi, A. Electrical Insulation Conference and Electrical
Manufacturing & Coil Winding Technology Conference, 2003. Proceedings, Vol., Iss.,
23-25 Sept. 2003. Pages: 535- 541
[ 5 ] Input current shaping in brushless DC motor drives utilizing inverter current
control. Skinner, J.; Lipo, T.A. Electrical Machines and Drives, 1991. Fifth
International Conference on (Conf. Publ. No. 341), Vol., Iss., 11-13 Sep 1991.
Pages:121-125
[ 6 ] Maxfield, Clive. The Design Warroirs’ s Guide to FPGAs. Newnes Publications, 2004. 1
– 22
[ 7 ] Altera Corporation. Quartus II Handbook, Volume 1. June 2004. Sections 6–1 to 6-16
[ 8 ] DARPA Grand Challenge. Rules. http://www.darpa.mil/grandchallenge. October 8th
2004. 1 – 31.
[ 9 ] HowStuffWorks, Inc. How Cruise Control Works. http://auto.howstuffworks.com/cruise-
control13.html. 1988-2004. 1-2
[ 10 ] Auto Week. 2002 Toyota Sequoia. http://www.autoweek.com?articleId=3383.
December 3rd 2001. 1-7
[ 11 ] CAN in Automation. Can Specifications 2.0, Part B.
[ 12 ] Modbus.org. MODBUS over Serial Lines Specification & Implementation guide.
http://modbus.org/ . December 2nd 2002. 1 - 44.
[ 13 ] Phillips Semiconductors. The I2C-Bus Specification Version 2.1. January 2001. 1-46.
[ 14 ] National Instruments. Serial Instrument Control. http://www.ni.com. 2004. 712 – 721.
[ 15 ] Lipowsky Idustrial-Elektronik. CAN-Tools Mikrocontroller. http://www.dgeinc.net.
2004. 1-5.
[ 16 ] Widrow, Bernard. Adaptive Signal Processing. Upper Saddle River, NJ: Prentice-Hall,
1985.