Imitation Learning for Generalizable Self-driving Policy with Sim-to-real Transfer

TL;DR

This paper explores how imitation learning combined with transfer learning can enable autonomous robots to learn lane-following in simulation and successfully transfer these skills to real-world environments, specifically in the Duckietown setting.

Contribution

It introduces and compares multiple imitation learning and sim-to-real transfer methods for autonomous lane-following, demonstrating their effectiveness in real-world deployment.

Findings

All methods successfully transferred to real-world Duckietown environment.

Transfer learning improved real-world performance over simulation-only models.

Different methods showed varying advantages in robustness and accuracy.

Abstract

Imitation Learning uses the demonstrations of an expert to uncover the optimal policy and it is suitable for real-world robotics tasks as well. In this case, however, the training of the agent is carried out in a simulation environment due to safety, economic and time constraints. Later, the agent is applied in the real-life domain using sim-to-real methods. In this paper, we apply Imitation Learning methods that solve a robotics task in a simulated environment and use transfer learning to apply these solutions in the real-world environment. Our task is set in the Duckietown environment, where the robotic agent has to follow the right lane based on the input images of a single forward-facing camera. We present three Imitation Learning and two sim-to-real methods capable of achieving this task. A detailed comparison is provided on these techniques to highlight their advantages and disadvantages.