Articulated Vehicle Docking Control With Nonlinear MPC

The patent implements a model predictive control (MPC) framework that continuously measures the actual state of the articulated vehicle and adjusts control inputs based on predicted future states. This closed-loop feedback mechanism compensates for modeling errors by comparing predicted trajectories with actual vehicle behavior and correcting deviations in real-time, thereby reducing sensitivity to inaccuracies in the n-trailer model while maintaining the ability to incorporate state constraints through explicit constraint formulation in the optimization problem.

The patent transitions from static linear MPC to dynamic nonlinear MPC that adapts to the changing operational modes of the articulated vehicle. The controller incorporates time-varying system matrices and constraints that change based on whether the vehicle is in forward or backward motion. This dynamic adaptation allows the same control framework to handle diverse maneuvers including straight-line cruising, curved path following, and reverse parking operations, thereby achieving versatility without sacrificing the systematic structure of MPC.

The patent employs a two-stage control approach where the first stage uses sampling-based motion planning to generate a coarse docking trajectory that brings the vehicle close to the target position. The second stage then refines this trajectory using model predictive control with tight state constraints that explicitly limit positioning errors. This preliminary action followed by refinement allows the automated system to achieve the extremely high precision required for safe docking while maintaining automation benefits, as the constraint formulation proactively prevents errors before they cause infrastructure damage.

The patent models the articulated vehicle as a segmented n-trailer system where each trailer is treated as a separate dynamic entity with its own state variables and control influences. The model predictive control framework formulates the overall tracking problem as a hierarchical optimization that independently considers each trailer segment while accounting for their coupled dynamics. This segmentation approach allows the system to handle disturbances on individual trailers by locally optimizing control inputs for affected segments, thereby maintaining reliable tracking performance even when many trailers are connected in series without active steering on the trailers themselves.