Sim-to-real Transfer of Deep Reinforcement Learning Agents for Online Coverage Path Planning

Sim-to-real transfer presents a difficult challenge, where models trained in simulation are to be deployed in the real world. The distribution shift between the two settings leads to biased representations of the perceived real-world environment, and thus to suboptimal predictions. In this work, we tackle the challenge of sim-to-real transfer of reinforcement learning (RL) agents for coverage path planning (CPP). In CPP, the task is for a robot to find a path that visits every point of a confined area. Specifically, we consider the case where the environment is unknown, and the agent needs to plan the path online while mapping the environment. We bridge the sim-to-real gap through a semi-virtual environment with a simulated sensor and obstacles, while including real robot kinematics and real-time aspects. We investigate what level of fine-tuning is needed for adapting to a realistic setting, comparing to an agent trained solely in simulation. We find that a high model inference frequency is sufficient for reducing the sim-to-real gap, while fine-tuning degrades performance initially. By training the model in simulation and deploying it at a high inference frequency, we transfer state-of-the-art results from simulation to the real domain, where direct learning would take in the order of weeks with manual interaction, i.e., would be completely infeasible.

Robust Predictive Motion Planning by Learning Obstacle Uncertainty

Safe motion planning for robotic systems in dynamic environments is nontrivial in the presence of uncertain obstacles, where estimation of obstacle uncertainties is crucial in predicting future motions of dynamic obstacles. The worst-case characterization gives a conservative uncertainty prediction and may result in infeasible motion planning for the ego robotic system. In this paper, an efficient, robust, and safe motion-planing algorithm is developed by learning the obstacle uncertainties online. More specifically, the unknown yet intended control set of obstacles is efficiently computed by solving a linear programming problem. The learned control set is used to compute forward reachable sets of obstacles that are less conservative than the worst-case prediction. Based on the forward prediction, a robust model predictive controller is designed to compute a safe reference trajectory for the ego robotic system that remains outside the reachable sets of obstacles over the prediction horizon. The method is applied to a car-like mobile robot in both simulations and hardware experiments to demonstrate its effectiveness.