SE(3) stochastic method improves 6-DOF spacecraft rendezvous safety

A new paper from Fabio D'Onofrio and Renato Zanetti, published on arXiv (2608.04114), presents an intrinsic stochastic successive convexification method formulated on the Special Euclidean group SE(3). Unlike conventional approaches that treat position and attitude separately or add stochastic dispersion only after a deterministic reference is fixed, this method jointly optimizes the nominal trajectory, covariance, and feedback law directly on the SE(3) manifold. This captures the intrinsic coupling between translational and rotational motion uncertainty — critical for rendezvous problems with safety constraints depending on full relative pose, such as collision avoidance, docking corridors, camera field of view, and probabilistic force and torque bounds.

In numerical simulations, the SE(3)-based method shapes closed-loop dispersion and improves probabilistic constraint satisfaction compared to tracking a deterministic reference with a feedback linearization controller. The work extends stochastic successive convexification — originally developed for Euclidean spaces — to nonlinear manifolds, enabling consistent covariance steering for rigid body pose trajectories. With 39 pages and 9 figures, the paper offers a rigorous framework for safer, more reliable spacecraft autonomy. It could directly impact future orbital rendezvous and docking missions, where accounting for uncertainty in both position and attitude is essential to meet stringent safety requirements under limited fuel and time.