Multi-Agent Imitation Learning for Driving Simulation

Bhattacharyya, Raunak P.; Phillips, D. C.; Wulfe, Blake; Morton, Jeremy R.; Kuefler, Alex; Kochenderfer, Mykel J.

doi:10.48550/arxiv.1803.01044

Cited by 1 publication

(6 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We compare Horizon GAIL to a number of baselines: BC, GAIL [4] and PS-GAIL [34], using the same dataset and observation and action spaces to train all methods. We show results using the best hyperparameters we found after tuning them separately for each method.…”

Section: Resultsmentioning

confidence: 99%

“…Related to ViBe are several existing LfD methods that learn road and pedestrian behaviour [29], [30], [31], [32]. Most relevant is learning highway merging behaviour [33], [34] from NGSIM [35], a publicly available dataset of vehicle trajectories. However, these methods again rely on manual labelling, synthetic data or specialised equipment to obtain the trajectories, while ViBe learns from raw, unlabelled videos of behaviour.…”

Section: B Learning From Demonstrationmentioning

confidence: 99%

“…Our simulator generates such a representation based on both the static and dynamic context. Pseudo-LiDAR readings, similar to those in [34], are used to represent different aspects of the static (e.g., zebra crossings and roads) and dynamic (e.g., distance and velocity of other agents) context of the agent. In addition, we provide information such as the agent's heading, distance from goal, and velocity.…”

Section: B Simulationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

“…In multi-agent situations, GAIL agents trained in a singleagent setting can fail to generalise to multi-agent settings [34]. PS-GAIL [34] is an extension to GAIL that addresses this issue by gradually increasing the number of agents controlled by the policy during training. We compare to PS-GAIL experimentally in Section V. However, it is complementary to the Horizon GAIL method we propose in Section IV-C and future work can focus on using them in conjunction.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Learning From Demonstration in the Wild

Behbahani

Shiarlis

Chen

et al. 2019

2019 International Conference on Robotics and Automation (ICRA)

View full text Add to dashboard Cite

Learning from demonstration (LfD) is useful in settings where hand-coding behaviour or a reward function is impractical. It has succeeded in a wide range of problems but typically relies on manually generated demonstrations or specially deployed sensors and has not generally been able to leverage the copious demonstrations available in the wild: those that capture behaviours that were occurring anyway using sensors that were already deployed for another purpose, e.g., traffic camera footage capturing demonstrations of natural behaviour of vehicles, cyclists, and pedestrians. We propose video to behaviour (ViBe), a new approach to learn models of behaviour from unlabelled raw video data of a traffic scene collected from a single, monocular, initially uncalibrated camera with ordinary resolution. Our approach calibrates the camera, detects relevant objects, tracks them through time, and uses the resulting trajectories to perform LfD, yielding models of naturalistic behaviour. We apply ViBe to raw videos of a traffic intersection and show that it can learn purely from videos, without additional expert knowledge.

show abstract

Section: Resultsmentioning

confidence: 99%