Automatically Learning Fallback Strategies with Model-Free Reinforcement Learning in Safety-Critical Driving Scenarios
This addresses safety and robustness for autonomous vehicles by enabling them to handle unexpected scenarios, though it is incremental as it builds on existing RL methods.
The paper tackles the problem of learning fallback strategies for autonomous vehicles in safety-critical driving scenarios using model-free reinforcement learning, and shows that the method can learn useful policies that would otherwise be missed during training.
When learning to behave in a stochastic environment where safety is critical, such as driving a vehicle in traffic, it is natural for human drivers to plan fallback strategies as a backup to use if ever there is an unexpected change in the environment. Knowing to expect the unexpected, and planning for such outcomes, increases our capability for being robust to unseen scenarios and may help prevent catastrophic failures. Control of Autonomous Vehicles (AVs) has a particular interest in knowing when and how to use fallback strategies in the interest of safety. Due to imperfect information available to an AV about its environment, it is important to have alternate strategies at the ready which might not have been deduced from the original training data distribution. In this paper we present a principled approach for a model-free Reinforcement Learning (RL) agent to capture multiple modes of behaviour in an environment. We introduce an extra pseudo-reward term to the reward model, to encourage exploration to areas of state-space different from areas privileged by the optimal policy. We base this reward term on a distance metric between the trajectories of agents, in order to force policies to focus on different areas of state-space than the initial exploring agent. Throughout the paper, we refer to this particular training paradigm as learning fallback strategies. We apply this method to an autonomous driving scenario, and show that we are able to learn useful policies that would have otherwise been missed out on during training, and unavailable to use when executing the control algorithm.