Intrinsic Stochastic Successive Convexification on SE(3) for Chance Constrained 6-DOF Rendezvous

What the math gives to ML

The paper's transferable mechanism is intrinsic stochastic successive convexification on a Lie group: represent pose increments in the se(3) tangent algebra, retract them through the exponential map, and optimize nominal motion, covariance, and feedback jointly under chance constraints. This avoids treating translation and rotation as independent Euclidean variables and provides a principled way to propagate uncertainty through coupled pose dynamics. The direct neural-network transfer is a pose-valued recurrent or world-model head whose hidden pose is updated by learned se(3) twists, with covariance steering and differentiable Gaussian chance penalties controlling long-horizon drift and safety.