A Bayesian Approach to Reinforcement Learning of Vision-Based Vehicular Control
This work addresses autonomous driving control for vehicles, but it is incremental as it builds on existing reinforcement learning and simulation frameworks.
The paper tackles vision-based autonomous driving by proposing a Bayesian reinforcement learning method that uses semantic segmentation maps from camera images, showing that training on ground truth data yields better performance than on estimated data, with reduced training time and superior benchmark results in the CARLA simulator.
In this paper, we present a state-of-the-art reinforcement learning method for autonomous driving. Our approach employs temporal difference learning in a Bayesian framework to learn vehicle control signals from sensor data. The agent has access to images from a forward facing camera, which are preprocessed to generate semantic segmentation maps. We trained our system using both ground truth and estimated semantic segmentation input. Based on our observations from a large set of experiments, we conclude that training the system on ground truth input data leads to better performance than training the system on estimated input even if estimated input is used for evaluation. The system is trained and evaluated in a realistic simulated urban environment using the CARLA simulator. The simulator also contains a benchmark that allows for comparing to other systems and methods. The required training time of the system is shown to be lower and the performance on the benchmark superior to competing approaches.