Researchers have posted a new paper on arXiv that investigates behavior cloning (BC) for autonomous driving in simulated urban environments. The approach uses a bird's-eye view (BEV) representation of the scene, and the model is trained via supervised learning to mimic expert driving behaviors.

The paper, titled 'Investigating Behavior Cloning from Few Demonstrations for Autonomous Driving based on Bird's-Eye View in Simulated Cities,' specifically addresses the challenge of learning from a small number of demonstrations. This is a practical concern because collecting large amounts of expert driving data is expensive and time-consuming.

The abstract describes the use of BC for this task but does not provide experimental details or results in the available text. The study is available on arXiv under identifier 2610.20203.