Efficient simulation framework brings advanced robotics to space operations
mjorbit: A Simulation Framework for Space Robotics
Robotics
Summary
Simulating robots in space is hard because they move around large, complex machines in weightless environments. The authors studied different ways to combine space movement with robot simulation. They made a new tool called mjorbit, which uses an existing robot simulator but adds space-specific physics and parts. They show how this tool can help control robots working on satellites or space stations using advanced methods like predictive control and learning. The tool and examples are available for anyone to use.
What this means in practice
- •For space robotics engineers: Simulate and test control strategies for robots performing assembly and servicing tasks in orbit using realistic spacecraft physics.
- •For robotics algorithm developers: Develop and evaluate advanced control methods like reinforcement learning for robots operating in spacecraft environments.
Authors
John Z. Zhang, Joris Verhagen, Fausto Vega, Patrick McKeen, Zachary Manchester
Abstract
This paper presents a general framework for simulating multi-body space robots with contact. We bring efficient, large-scale robot simulation to in-space servicing, assembly, and manufacturing applications. First, we perform an empirical trade study of methods for coupling orbit propagation with existing robotics simulation frameworks. Next, we present mjorbit, a general, flexible, and performant framework built on the MuJoCo engine widely used in robotics, to which we add key spacecraft dynamics, actuators, and sensors. We provide a low-latency C++ CPU backend and a high-throughput GPU backend behind a simple Python API. We demonstrate mjorbit by solving several realistic on-orbit case studies with both model-predictive control and reinforcement learning. Open-source code and examples are available at: https://johnzhang3.github.io/mjorbit/