safe-control-gym: a Unified Benchmark Suite for Safe Learning-based Control and Reinforcement Learning in Robotics

TL;DR

This paper introduces safe-control-gym, an open-source benchmark suite for evaluating and comparing safe learning-based control and reinforcement learning methods in robotics, supporting multiple systems and tasks with standardized API extensions.

Contribution

The paper presents a unified benchmark suite that enables equitable comparison of control and reinforcement learning approaches in robotics, with extended API features for safety and disturbance simulation.

Findings

Demonstrated the use of safe-control-gym to compare different control methods.

Showcased the platform's ability to evaluate safety, data efficiency, and performance.

Provided implementations for multiple robotic systems and tasks.

Abstract

In recent years, both reinforcement learning and learning-based control -- as well as the study of their safety, which is crucial for deployment in real-world robots -- have gained significant traction. However, to adequately gauge the progress and applicability of new results, we need the tools to equitably compare the approaches proposed by the controls and reinforcement learning communities. Here, we propose a new open-source benchmark suite, called safe-control-gym, supporting both model-based and data-based control techniques. We provide implementations for three dynamic systems -- the cart-pole, the 1D, and 2D quadrotor -- and two control tasks -- stabilization and trajectory tracking. We propose to extend OpenAI's Gym API -- the de facto standard in reinforcement learning research -- with (i) the ability to specify (and query) symbolic dynamics and (ii) constraints, and (iii) (repeatably) inject simulated disturbances in the control inputs, state measurements, and inertial properties. To demonstrate our proposal and in an attempt to bring research communities closer together, we show how to use safe-control-gym to quantitatively compare the control performance, data efficiency, and safety of multiple approaches from the fields of traditional control, learning-based control, and reinforcement learning.