Collision-aware trajectory planning for passenger boarding bridges in airport apron

Passenger Boarding Bridges (PBBs) are the primary physical interface between terminal buildings and aircraft, yet current semi-automated systems execute pre-programmed trajectory sequences designed for static environments and lack real-time obstacle awareness. This makes them unsuitable for aprons, where ground support vehicles, cargo equipment, and personnel continuously occupy the docking corridor. This study presents and validates a collision-aware trajectory planning framework for autonomous PBB docking, implemented in Robot Operating System 2 (ROS 2) and MoveIt 2. Rather than committing to a single fixed path, the framework uses the MoveIt Task Constructor (MTC) to evaluate four topologically distinct docking routes in parallel and autonomously select the fastest collision-free trajectory, falling back automatically when the primary route is obstructed. The framework was validated in kinematic simulation over 80 runs, comparing RRTConnect and RRT* across unobstructed and obstructed scenarios, with obstacles registered directly as collision objects. A planner returning a trajectory proved an insufficient safety criterion: under obstruction, RRTConnect returned a trajectory in every run, but 70% violated joint constraints or penetrated obstacle geometry at execution-time validation, whereas RRT* produced no such failures across 40 runs and selected the correct fallback strategy in all 20 obstructed runs. RRT* required roughly 10 times more planning time than RRTConnect (21.3 s versus 1.7 s) yet yielded an identical execution time in the unobstructed scenario (96.3 s). These results indicate that asymptotically optimal planning and execution-time trajectory validation are both necessary for safe autonomous PBB docking under the tested single-obstacle configuration.