Print Email Facebook Twitter Learning from Demonstrations of Critical Driving Behaviours Using Driver’s Risk Field Title Learning from Demonstrations of Critical Driving Behaviours Using Driver’s Risk Field Author DU, YURUI (TU Delft Mechanical, Maritime and Materials Engineering; Siemens Digital Industries Software) Contributor Kober, J. (mentor) Tong, Son (graduation committee) Acerbo, Flavia (graduation committee) Degree granting institution Delft University of Technology Programme Mechanical Engineering Date 2022-10-20 Abstract In recent years, imitation learning (IL) has been widely used in industry as the core of autonomous vehicle (AV) planning modules. However, previous work on IL planners shows sample inefficiency and low generalisation in safety-critical scenarios, on which they are rarely tested. As a result, IL planners can reach a performance plateau where adding more training data ceases to improve the learnt policy. First, our work presents an IL model using spline coefficient parameterisation and offline expert queries to enhance safety and training efficiency. Then, we expose the weakness of the learnt IL policy by synthetically generating critical scenarios through optimisation of parameters of the driver's risk field (DRF), a parametric human driving behaviour model implemented in a multi-agent traffic simulator based on the Lyft Prediction Dataset. To continuously improve the learnt policy, we retrain the IL model with augmented data. Thanks to the expressivity and interpretability of the DRF, the desired driving behaviours can be encoded and aggregated to the original training data. Our work constitutes a full development cycle that can efficiently and continuously improve the learnt IL policies in closed-loop. Finally, we show that our IL planner developed with 30 times less training resource still has superior performance compared to the previous state-of-the-art. Subject imitation learningautonomous drivingcritical scenario generationmodel-based multi-agent simulator To reference this document use: http://resolver.tudelft.nl/uuid:dd4e6fd4-73c8-4f4d-82a6-adcf2d5db98d Part of collection Student theses Document type master thesis Rights © 2022 YURUI DU Files PDF TUD_Report_YuruiDu.pdf 7.3 MB Close viewer /islandora/object/uuid:dd4e6fd4-73c8-4f4d-82a6-adcf2d5db98d/datastream/OBJ/view