A unified framework for compliant control and trajectory planning in robotic in-orbit assembly

Robotic In-Orbit Assembly (R-IOA) is a key technology for the construction of large-scale space infrastructure. A significant challenge in this field is the precise docking of modular components, particularly those with complex interfaces. To address this, safety must be considered at both the control and planning levels. At the control level, traditional compliant control strategies show poor adaptability to complex contact and often face challenges in explicitly enforcing physical limitations, such as joint position and torque bounds. Furthermore, while compliant control can effectively manage steady-state contact forces, it is fundamentally limited in mitigating the initial transient impact. This peak force is predominantly determined by the pre-impact velocity, a parameter dictated by the trajectory planner. Conventional planners, however, often generate dynamically infeasible trajectories using decoupled methods that violate the kinematic consistency of rigid-body motion, and they are generally limited in their ability to strictly enforce velocity constraints. To address these dual challenges through an integrated approach, this paper proposes a unified safety framework. This framework combines comprehensive enhancements at both the planning and control layers. At the planning level, a time-optimal trajectory generator operating on the SE(3) manifold produces motions that are dynamically feasible by construction. This ensures that velocity constraints are strictly enforced to proactively minimize impact forces and that the trajectory respects the natural kinematics of rigid-body motion. At the control level, we introduce a Hierarchical Quadratic Programming based Adaptive Controller (HQP-AC). It reformulates compliant interaction as a constrained optimization problem to guarantee the strict enforcement of all hardware safety limits while adaptively managing the steady-state interaction. The effectiveness of the proposed approach was demonstrated through simulations of a representative docking scenario. Compared to a classical impedance controller with decoupled trajectory planning, the proposed framework reduces peak axial contact forces by 49% and steady-state contact forces by 30%, and successfully prevents the catastrophic joint limit violations observed in the baseline method. Furthermore, it achieves a final lateral position error of 0.18 mm and an orientation error of 1.08°, representing a significant improvement in docking accuracy.