Autonomous spacecraft inspection improves safety with chance constraints
Chance-Constrained Inspection Viewpoint Sequencing for On-Orbit Servicing via Safe Reinforcement Learning
Artificial Intelligence
Summary
Planning safe close-up inspections of spacecraft in orbit is tricky because you have to cover the whole target while saving fuel and avoiding collisions. The authors treat this task as a decision problem with safety limits and use a type of machine learning to find the best sequence of camera positions. Their method adapts the plans based on the risk of errors in navigation and movements, making safety violations rare and predictable. Tests showed it works better than standard methods that use fixed safety buffers, balancing coverage, fuel use, and collision risk.
What this means in practice
- •For space mission planners: Plan spacecraft inspection sequences balancing coverage, fuel, and safety with controllable collision risk for on-orbit servicing missions.
- •For satellite operators: Integrate adaptive viewpoint selection into autonomous inspection to improve safety and efficiency during in-orbit satellite servicing.
Authors
Chloé Carer, Adam Abdin
Abstract
Autonomous inspection of a target spacecraft is typically the initial step in on-orbit servicing. Planning close-proximity observation viewpoints involves a complex combinatorial problem constrained by coverage, fuel, and safety considerations. This paper models the viewpoint selection as a Constrained Markov Decision Process (CMDP) and applies Proximal Policy Optimization with Lagrangian relaxation to solve it. The system combines a high-level planner, which sequences camera viewpoints on an observation sphere around the target, with a low-level model that assesses each transfer using the Clohessy--Wiltshire (CW) equations for relative motion. The deterministic constraints on single-impulse reachability and $Δv$ cap are enforced directly on the action set. Safety margins are affected by navigation and manoeuvre errors, making the actual clearance stochastic. To address this, a chance constraint on the realised trajectory is implemented, with a Lagrangian multiplier governed by PID control to adapt the penalty during training. The approach is tested on a Galileo-class navigation satellite in medium Earth orbit, modelled as a point cloud where visibility from each viewpoint considers field-of-view and line-of-sight occlusion. Across various risk budgets, the trained policy maintains the keep-out violation rate within the specified limit, achieving coverage close to its maximum and staying within the mission $Δv$ budget. Compared to the usual practice of applying a fixed margin to the keep-out radius and planning deterministically, the chance-constrained policy attains less than half the violation rate at comparable coverage. This makes the collision risk an explicit design parameter, eliminating the need for re-tuning per case. The learned policy selects a viewpoint without the repeated Monte Carlo risk estimates required by a greedy, chance-constrained planner at every decision point.