Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

31
Towards Autonomous Unmanned Aircraft for Low-Altitude Remote Sensing Torsten Merz Unmanned Aerial Vehicles in the Resources Industry – Perth, 26-27 June 2014 Autonomous Systems Program – CSIRO Computational Informatics www.csiro.au

description

Torsten Merz, Senior Research Scientist, CSIRO delivered the presentation at the 2014 Unmanned Aerial Vehicles (UAV) in the Resources Industry. The 2014 Unmanned Aerial Vehicles (UAV) in the Resources Industry explored the enormous potential of UAVs within mining and resources operations. For more information about the event, please visit: http://www.informa.com.au/uavresourcesconference14

Transcript of Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

Page 1: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

Towards Autonomous Unmanned Aircraft forLow-Altitude Remote SensingTorsten Merz

Unmanned Aerial Vehicles in the Resources Industry – Perth, 26-27 June 2014

Autonomous Systems Program – CSIRO Computational Informaticswww.csiro.au

Page 2: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 2 of 31

Overview

• Definitions, concepts and incentives• Aircraft system selection and operational considerations• Remote sensing workflow• Prototype systems developed at CSIRO• Autonomous aircraft and regulations• Conclusion

Page 3: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 3 of 31

Definitions

• Remote sensing: ”Remote sensing is a way of collecting andanalysing data to get information about an object without theinstrument used to collect the data being in direct contact withthe object.” [ESA]• Autonomous unmanned aircraft: ”An aircraft that does not re-

quire a human pilot to fly a given mission in a specified environ-ment.”• Low-altitude: airborne to a height manned aircraft typically don’t

cruise (<400ft AGL)

Page 4: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 4 of 31

Unmanned Aircraft Terminology

• UAV/S→RPA(S): Remotely Piloted Aircraft (System) [CASA/ICAO]• Drone: stingless male bee making no honey; term introduced

by the military for an unmanned aircraft used as training target• Autonomous aircraft: ”An unmanned aircraft that does not allow

pilot intervention in the management of flight” [CASA/ICAO]• Remote pilot: ”The person who manages the flight controls of a

remotely piloted aircraft during flight time.” [CASA/ICAO]

Page 5: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 5 of 31

Why Remote Sensing?

• Coverage of large (possibly difficult to access) areas• High temporal and spatial resolution• Lower costs, faster and safer compared to ground-based inves-

tigation/monitoring• Low-altitude:

– typically higher temporal and spatial resolution– possibly lower costs (operations, sensors)– easier from a regulatory perspective– BUT coverage of large areas problematic → swarm of un-

manned aircraft

Page 6: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 6 of 31

Applications in the Resources Industry

• Structural geology• Mineral identification• Regolith mapping• Geophysical surveys• Site scoping and mapping• Stockpile volume mapping• Stope mapping• Inspection of fixed plant and to get a better analysis of potential shutdown

for repair• Monitor for any subsidence above or adjacent to underground mines• Thermal monitoring of coal stockpiles• Pit mapping for highwalls, pit walls and development faces• Surveying and monitoring offshore oil/gas infrastructure• New uses for aerial real time truck management, site and remote infras-

tructure monitoring, and machinery tracking• Remove risk and increase safety for site personnel• Monitor the performance of mine tailings facilities

Page 7: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 7 of 31

Autonomy

• No clear definition• Autonomy is regarded as a relational notion• Relative autonomy: relation between a system, its human op-

erator, the mission and the environment (ALFUS framework,NIST)• Level of autonomy can be related to human independence, mis-

sion complexity and environmental complexity• We do not consider self-motivated systems here• Full autonomy→ no human intervention required to complete a

given mission• Automation can be seen as low-level autonomy• Autonomy does not necessarily imply dependability

Page 8: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 8 of 31

Dependability

• Term introduced by Jean-Claude Laprie in the early 80s forcomputing systems• ”ability to deliver service that can justifiably be trusted”• Three parts: attributes of dependability, threats to dependability,

means to attain dependability• Attributes: availability, reliability, safety, confidentiality, integrity,

maintainability• Threats: faults, errors, failures• Means: fault prevention, fault tolerance, fault removal, fault fore-

casting• Our research is towards dependable autonomy of unmanned

aircraft systems

Page 9: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 9 of 31

Why Autonomous Unmanned Aircraft?

• Autonomous aircraft can consistently perform flights which aredifficult for human pilots→ better remote sensing data• Costs and availability argument• Autonomy→ relaxed communication requirements• Facilitates operations of swarms of unmanned aircraft• Human piloted unmanned aircraft not necessarily safer:

– situational awareness and control authority in case of fail-ures typically limited for pilots of unmanned aircraft com-pared to manned aircraft

– human errors– human machine interaction prone to problems

Page 10: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 10 of 31

Aircraft System Selection

• Rotorcraft vs. fixed-wing vs. aerostats• Type depends on remote sensing task and type of operation:

– flight envelope (hover requirement?)– payload capacity and endurance– propulsion (combustion/electric)– available area for takeoff and landing– handling (transport, piloting)

• Size/gross weight of aircraft• Costs, dependability, certification

Page 11: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 11 of 31

Type of Operations

• Interactive vs. non-interactive operations (→ level of autonomy)• Operations from fixed base locations vs. operations from ran-

dom locations• Tethered vs. un-tethered flight• Single aircraft vs. swarm• COTS solution vs. custom development: fully integrated sys-

tems optimised for a specific tasks often not available→ com-promised solution

Page 12: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 12 of 31

Remote sensing workflow

Pre-mission (mission specification)• mission area definition• sensor selection/configuration• flight plan creation with image capture positions incorporating

resolution, coverage, timing requirementsMission executionPost-mission• check data (image quality, on-site)• check coverage (on-site)• data analysis

Page 13: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 13 of 31

Dependable Autonomous Flight Services

• Take off• Hover (including heading change)• Straight line and circular curve path following• Waypoint flight with obstacle avoidance• (Waypoint flight with air traffic control interaction)• Terrain following• Close-range inspection• Duck manoeuver (aircraft avoidance)• (Landing)

(CSIRO single-rotor research helicopters)

Page 14: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 14 of 31

Infrastructure Inspection Helicopter Prototype

Inspection camera (forward−looking)2D LIDAR (vertical scan)

• 12.3kg gross weight• 1.78m rotor diameter• ∼1h endurance• Petrol engine

Smart Skies Project (www.arcaa.net/research/smart-skies-project)

Page 15: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 15 of 31

Infrastructure Inspection Mission

1.4km

Page 16: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 16 of 31

Phenocopter

Remote sensing cameras (downward−looking)

Page 17: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 17 of 31

Mission Planning Tools

• Tools for missions in flat (shown above) and mountainous terrain• Platform independent software (Linux, Windows, Mac OS)

Page 18: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 18 of 31

Setup

• Helicopter + wireless operator interface + transport vehicle• Takeoff area: ∼4 m diameter, even ground, <5 degrees slope• Operation: 1-2 people, <30 min setup time• 4 helicopters, 55 days, 37.5 flight hours

Page 19: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 19 of 31

Typical Crop Monitoring Mission

50m

Autonomous flight

Ground station

First waypoint

Last waypoint

landingand

Take offManual flight

[m]

[s]

Autonomous flight

Manual flight

Height

Reference

Page 20: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 20 of 31

Post-flight Analysis Tools

• Visualisation of camera poses and quick access of correspond-ing images recorded during flight• Automated image analysis

Page 21: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 21 of 31

Wheat Trial - Multispectral Imaging

• Visual, NIR, LWIR spectral range• Separate calibrated cameras• All images recorded on-board in raw format with camera pose

Page 22: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 22 of 31

Wheat Trial - Image Analysis

• Plant coverage estimation• DEM for lodging analysis• Contact: Scott Chapman, CSIRO Plant Industry

Page 23: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 23 of 31

Weed Survey Helicopter Prototype

Pilotless low-altitude flight in mountainous terrainProject ResQu (www.arcaa.net/research/resqu)

Page 24: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 24 of 31

Video - Miconia Weed Survey

El Arish near Cairns

Page 25: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 25 of 31

Unmanned Rotorcraft for Sponcom Monitoring

• Thermal infrared imagery on-board a quadcopter• Enables automatic mapping and monitoring of large areas for

heatings• Field trials at old coal mine workings• Contact: John Malos, CSIRO Earth Science and Resource En-

gineering (CESRE)

Page 26: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 26 of 31

Unmanned Rotorcraft for Surface Characterisation

• Acquire high resolution spatial data (∼100µm) of fractures inslope or mine bench• Enables estimation of surface roughness for geotechnical and

hydrogeological analysis• Contact: Peter Dean, CSIRO Earth Science and Resource En-

gineering (CESRE)

Page 27: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 27 of 31

3D Mapping

• CSIRO Zebedee 3D mapping technology integrated on a quad-copter• Faster mapping and maps with less gaps• Demonstrated at a marble quarry in NSW• Contact: Robert Zlot, CSIRO Computational Informatics (CCI)

Page 28: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 28 of 31

Autonomous Systems and Regulations

• CASA/ICAO currently have different understanding of meaningof autonomy• Current and proposed new regulations don’t consider opera-

tions of pilotless aircraft in Australia• Pilotless does not mean unsupervised• It used to be feasible to legally conduct R&D work on pilotless

aircraft in Australia• We proposed a way to safely operate pilotless aircraft based on

scientific evidence• Current focus on certification of remote pilots and organisations

rather than autonomous systems↔ autonomous cars in the US• This may hinder the deployment of unmanned aircraft in the re-

sources industry especially considering future automation pro-grams

Page 29: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 29 of 31

RPAs and Regulations

• RPAs may include autonomous system components but a re-mote pilot must be able to manage the flight controls• Commercial operations: RPA operator certificate + certified re-

mote pilots• Less restrictions for operations of RPAs:

– with gross weight <150/100kg (2kg)– operations in Class G airspace <400ft AGL >3NM from

aerodromes– clear of populous areas– in daylight VMC conditions– within line-of-sight

Page 30: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

informa: Slide 30 of 31

Conclusion

• Unmanned aircraft useful for the resources industry pending fu-ture regulatory restrictions• RPAs commercially available and maturing but dependable pi-

lotless aircraft still in their infancy• Carefully choose systems and how to operate• Integrated systems optimised for specific tasks may not be avail-

able off-the-shelf• Provided incentives for autonomous aircraft and discussed reg-

ulatory restrictions• Pilotless aircraft technology is being developed in CSIRO and

validated in real-world applications

Page 31: Torsten Merz - CSIRO - Towards autonomous unmanned aircraft for low-altitude remote sensing

Autonomous Systems Program – CSIRO Computational Informaticswww.csiro.au

CCI Autonomous Systems Program

Dr.-Ing. Torsten Merzt +61 7 3327 4123e [email protected] http://www.csiro.au