SummerCourse-France-2018-Presentation-v0.3 - Modo de...
Transcript of SummerCourse-France-2018-Presentation-v0.3 - Modo de...
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018
Intelligent Interaction between Vehicles and VRUs
Miguel Ángel Sotelo
SPAIN
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 2
Motivation
Proposed approach
Pedestrian pose measurement
Dimensionality reduction and Prediction
Experimental results
Conclusions and Future Work
Content
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 3
Motivation
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 4
Motivation
• Pedestrian Path Prediction in the Automotive:
– Further improvement in state-of-the-art ADAS by meansof action classification based on body language reading
– Walking, Stopping, Starting, Bending-in
– Improvement of accuracy in 30-50 cm:
– Difference between effective and non-effectiveintervention in emergency braking systems
– Initiation of emergency braking 0.16 s in advance canpotentially reduce severity of accidents injuries by 50%
– Early recognition of pedestrians stopping/walking actionscan provide more accurate last-second active interventions
Strong gains are expected in the performance and reliability of active
pedestrian protection systems
Strong gains are expected in the performance and reliability of active
pedestrian protection systems
Strong gains are expected in the performance and reliability of active
pedestrian protection systems
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 5
Motivation
Proposed approach
Pedestrian pose measurement
Dimensionality reduction and Prediction
Experimental results
Conclusions and Future Work
Content
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 6
Proposed Approach
Global Scheme
Off-line Motion Capture System
Recovery of 3D pose and position
Lateral predicted position
On-line
Probabilistic Training
Knowledge of
Pedestrians
dynamics
Stereo Cameras
Transformation to latent space +
prediction
Geometric processing
3D Pedestrian Pose
Estimation
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 7
Motivation
Proposed approach
Pedestrian pose measurement
Dimensionality reduction and Prediction
Experimental results
Conclusions and Future Work
Content
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 8
Pedestrian Pose Measurement
Pedestrian Skeleton considered in this research
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 9
Pedestrian Pose Measurement
Body parts detection using Deep Learning
St: Confidence MapsLt: Parts Affinity Maps (PAF) (Z. Cao. CVPR 2017)
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 10
Pedestrian Pose Measurement
Body parts detection using Deep Learning
Confidence maps of the right wrist (first row) and PAFs(second row) of right forearm across stages
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 11
Pedestrian Pose Measurement
Body parts detection using Deep Learning
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 12
Pedestrian Pose Measurement
Body parts detection using Deep Learning
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 13
Motivation
Proposed approach
Pedestrian pose measurement
Dimensionality reduction and Prediction
Experimental results
Conclusions and Future Work
Content
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 14
Dimensionality Reduction
Major Goal
Learning the pedestrians motion dynamics by reducingthe dimensionality of input data and to make it moremanageable and interpretable in a low dimensional latentembedding
Method considered
GPDM (Gaussian Process with Dynamical Model)
Input data-set
CMU (41 body joints obtained with motion capture syst.)
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 15
Dimensionality Reduction
CMU Sequence Example
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 16
Dimensionality Reduction
GPDM (1/3)
1. Nonlinear probabilistic generalization of PCA: marginalization of the mapping (W) and optimization of the latent variables (X)
2. Conditional density in GPDM for a centered observed data Y, latent variable X, and kernel hyperparameters θ
where W is a scaling matrix and K is a kernel matrix with:
where δi,j is the Kronecker delta function
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 17
Dimensionality Reduction
GPDM (2/3)
3. Dynamic mapping from latent coordinates:
where Xout=[X2, … , XN], d is the model dimension and KX is the kernel built from {X1, … , XN-1} with:
where βi are hyperparameters
4. Minimization of –ln p(X,θ,β,W|Y) using SCG
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 18
Dimensionality Reduction
GPDM (3/3)
Prediction
1. Predicted latent position in the next frame (kX is a column vector)
2. Reconstructed 3D pose (kY is a column vector)
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 19
Learning Pedestrian Motion
Pedestrian walking 6 steps (using B-GPDM)
Green: projection of pedestrian motion sequence onto the subspace.
Variance in color code: cold (small) to warm (big)
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 20
Experimental Results
Video sequence showing prediction results
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 21
Motivation
Proposed approach
Pedestrian pose measurement
Dimensionality reduction and Prediction
Experimental results
Conclusions and Future Work
Content
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 22
General Method - Overview
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 23
Datasets
Frame Rate of training data
CMU data-set: 120 fps (frames per second)
Feature Vector
UAH: 11 body joints (3D pose + 3D motion: 11x6)
CMU: 41 (we consider only the 11 joints in common with UAH)
Pedestrian Behaviors
Walking (toward the curb to enter the road lane)
Stopping (at the curb)
Starting (from the curb)
Standing
Prediction Horizon
Up to 1.0 s
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 24
Experimental Results
UAH Dataset - Breakdown
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 25
Manual Labelling of Pedestrian Poses
Standing-Starting Transition
Criteria: movement of leg forward
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 26
Manual Labelling of Pedestrian Poses
Starting-Walking Transition
Criteria: pedestrian places foot on the ground after one stride
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 27
Manual Labelling of Pedestrian Poses
Walking-Stopping Transition
Criteria: pedestrian places foot on the ground in the final stride before full stop
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 28
Manual Labelling of Pedestrian Poses
Stopping-Standing Transition
Criteria: pedestrian comes to a full stop
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 29
Activity Recognition
Clues for prediction of “Starting” behaviors
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 30
Activity Recognition
Clues for prediction of “Stopping” behaviors
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 31
Activity Recognition
Pedestrian activity modelled as HMM
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 32
Activity Recognition
Pedestrian activity modelled as HMM
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 33
Activity Recognition
Pedestrian activity modelled as HMM
Prior
Emission (Likelihood)
α: SSE for pedestrian poseβ: SSE for pedestrian displacement
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 34
Activity Recognition
Recognition results for different features and #joints
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 35
Experimental Results
Probabilistic Action Classification
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 36
Experimental Results
Detection Delay - Summary
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 37
Experimental Results
UAH Prototype vehicle - DRIVERTIVE
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 38
Experimental Results
Experiments in UTAC proving ground (France) – BRAVE H2020 Project
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 39
Experimental Results
Experiments in UTAC proving ground (France) – BRAVE H2020 Project
Expected Contributions to Euro NCAP
- Recommendations to Euro NCAP WG on AD on how to measure performance of VRU prediction systems
- Enhanced performance in Euro NCAP tests by including predictions
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 40
Experimental Results
Experiments in UTAC proving ground (France) – BRAVE H2020 Project
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 41
Experimental Results
Experiments in UTAC proving ground (France) – BRAVE H2020 Project
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 42
Intelligent Interface with VRUs
GRAIL – GReen Assistant Interface Light
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018
Use Case 1 – Pedestrians (GRAIL-1)
The pedestrian stops and looks at the ego-vehicle
Ego-vehicle decelerates until full stop and indicates to thepedestrian that s/he can cross.
The pedestrian continues and completely crosses.
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018
Use Case 1 – Pedestrians (GRAIL-2)
The ego-vehicle decelerates, the pedestrian turns his/her head to the left indicating a possible intention to cross
The ego-vehicle comes to a full stop while indicating the pedestrian that s/he can cross.
The pedestrian crosses.
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018
Use Case 2 – Pedestrians (passing-by)
Pedestrian walking along the road
The ego-vehicle decelerates
Passes the pedestrian slowly.
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018
Use Case 3 – Pedestrians (avoidance)
25%
Crossing scenario with impact point at 25%.
AES activation to avoid the impact.
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018
Use Case 4 – Cyclists (passing-by)
It’s possible to overtake so deceleration takes place
and overtaking with the correct distance with the bicycle.
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018
Use Case 5 – Cyclists (traffic context)
The ego-vehicle decelerates and waits to the oncoming traffic to pass
After that it overtakes.
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 49
Intelligent Interface with VRUs
GRAIL – GReen Assistant Interface Light
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 50
Intelligent Interface with VRUs
GRAIL will be demonstrated at IEEE IROS 2018
First week of October, Madrid
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018 51
Motivation
Proposed approach
Pedestrian pose measurement
Dimensionality reduction and Prediction
Experimental results
Conclusions and Future Work
Content
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018
Conclusions and Future Work
Conclusions
- Accurate path pedestrian prediction is possible up to 1.0 s using body language traits and probabilistic dimensionality reduction techniques.
- Predictions are carried out in the latent space, yielding results with a potential accuracy of 15-20 cm at a time horizon of 1.0 s
- The generative solution is not scalable to thousands of examples. A new approach is needed for massive learning.
Future Research Challenges
- Taking a discriminative approach to the prediction problem: Deep Learning? Recurrent Neural Networks?
- VRU recognition (standing pedestrian, seating pedestrian, cyclist, motorcyclist).
- Context-based action prediction:
- Gaze direction, group behavior, type of road.
- Use of Probabilistic Graphical Models (Bayesian Networks).
Summer School on Cooperative Interacting Automobiles, Loire Valley, France. September 2018
Merci beaucoup!
Miguel Ángel Sotelo
[email protected] University of Alcalá, SPAIN
Intelligent Interaction between Vehicles and VRUs