Safe Tracking Control of an Underactuated Suspended Backpack via BLF-Based Backstepping and a Disturbance Observer

Active suspended backpacks offer potential for reducing metabolic cost and improving load-carrying efficiency during locomotion. However, gait variability and unmodeled human-robot interactions introduce disturbances and uncertainties that may cause load swing or structural collisions, resulting in high impact forces and safety risks. This paper addresses the safe trajectory tracking problem of an underactuated suspended backpack system modeled with a series elastic actuator (SEA), where the control input and target state lie on different channels. To estimate external disturbances and modeling errors in real time, a high-order disturbance observer (HODOB) is developed. A smooth backstepping controller incorporating a barrier Lyapunov function (BLF) is then designed to enforce state constraints and guarantee safe tracking. The resulting control law guarantees that the system state remains within predefined safety bounds while maintaining robustness against estimation errors and unmodeled dynamics. Simulation results confirm that the proposed method achieves accurate tracking while satisfying safety constraints, effectively rejecting disturbances, and reducing shoulder interaction across a range of load and walking conditions.