New AI method lets robots measure object inertia in real-time space missions

A team of researchers has developed a new method to estimate the inertia of objects grasped by robots in real time. The breakthrough is particularly useful for free-floating robotic systems operating in space. Accurate inertia estimation is vital for precise manipulation, especially in weightless environments where robot movements can alter overall positioning.

The method was created by Akiyoshi Uchida, Antonine Richard, Kentaro Uno, Miguel Olivares-Mendez, and Kazuya Yoshida. It builds on existing identification techniques but adds momentum conservation principles to work with unanchored robots in orbit. By using Recursive Least Squares with a novel regularisation term—Log-Determinant Divergence—the team improved both accuracy and stability in parameter estimation.

Testing has already taken place in simulations using the Mujoco physics engine. The results showed accurate identification of inertia parameters in various scenarios. Real-world experiments on a robotic platform are now planned to further validate the approach. The researchers also ensured the estimated parameters remain physically realistic. For example, the method maintains a positive definite inertia tensor, preventing impossible values. Future steps include comparing its performance against the Extended Kalman Filter and assessing its computational demands, noise sensitivity, and handling of complex object shapes. Looking ahead, the team aims to optimise robot trajectories for better estimation accuracy. They also plan to integrate the framework into an adaptive control system. Missions like ESA's ClearSpace-1 and NASA's OSAM-1 could benefit significantly, as both involve free-floating robots capturing uncooperative space debris.

This method addresses a key challenge in space robotics: dynamics-aware manipulation in orbital environments. By providing real-time inertia estimation, it supports safer and more precise operations for free-floating systems. Further testing and refinements will determine its full potential for upcoming space missions.