Co-learning Planning and Control Policies Constrained by Differentiable Logic Specifications

TL;DR

This paper introduces a reinforcement learning method that co-learns planning and control policies constrained by differentiable logic, significantly reducing training data needs and enabling complex, high-dimensional robot navigation tasks.

Contribution

It presents a novel RL approach that integrates differentiable logic specifications into co-learning planning and control, improving efficiency and scalability for complex robotic tasks.

Findings

Reduces sample complexity in training policies

Enables extraction of complex specifications from map images

Demonstrates effectiveness on simulated and real robots

Abstract

Synthesizing planning and control policies in robotics is a fundamental task, further complicated by factors such as complex logic specifications and high-dimensional robot dynamics. This paper presents a novel reinforcement learning approach to solving high-dimensional robot navigation tasks with complex logic specifications by co-learning planning and control policies. Notably, this approach significantly reduces the sample complexity in training, allowing us to train high-quality policies with much fewer samples compared to existing reinforcement learning algorithms. In addition, our methodology streamlines complex specification extraction from map images and enables the efficient generation of long-horizon robot motion paths across different map layouts. Moreover, our approach also demonstrates capabilities for high-dimensional control and avoiding suboptimal policies via policy alignment. The efficacy of our approach is demonstrated through experiments involving simulated high-dimensional quadruped robot dynamics and a real-world differential drive robot (TurtleBot3) under different types of task specifications.