Model-Predictive Policy Learning with Uncertainty Regularization for Driving in Dense Traffic
This addresses the problem of safe and efficient autonomous driving in dense traffic for robotics and AI applications, representing an incremental improvement by integrating uncertainty regularization into model-predictive learning.
The paper tackled the challenge of learning policies from observational data by proposing a method that unrolls a learned dynamics model and penalizes both the original cost and an uncertainty cost based on model predictions, achieving effective driving policies from traffic camera data without environment interaction.
Learning a policy using only observational data is challenging because the distribution of states it induces at execution time may differ from the distribution observed during training. We propose to train a policy by unrolling a learned model of the environment dynamics over multiple time steps while explicitly penalizing two costs: the original cost the policy seeks to optimize, and an uncertainty cost which represents its divergence from the states it is trained on. We measure this second cost by using the uncertainty of the dynamics model about its own predictions, using recent ideas from uncertainty estimation for deep networks. We evaluate our approach using a large-scale observational dataset of driving behavior recorded from traffic cameras, and show that we are able to learn effective driving policies from purely observational data, with no environment interaction.