Trajectory-Based Maneuvering Control for Autonomous Driving Comfort

The patent implements dynamic adaptation by enabling the manoeuvring controller to automatically adjust its configuration parameters in real-time based on the planned trajectory characteristics. The controller transitions from a static fixed configuration to a dynamic adaptive system that selects optimal parameters (such as PID gains, control horizon, or model predictive control settings) according to the current driving situation encoded in the trajectory plan, thereby achieving adaptability without requiring complex external sensor processing.

The patent changes the operational parameters of the manoeuvring controller based on trajectory characteristics. Specifically, it adjusts controller parameters (such as proportional, integral, and derivative gains in PID control, or weighting factors in model predictive control) in response to trajectory features like curvature, speed variations, and lane change magnitude. This parameter adaptation allows the same controller hardware to optimize performance across diverse driving situations without physical reconfiguration.

The patent extracts and utilizes only the essential trajectory characteristics from the trajectory plan that are sufficient for configuration selection, rather than processing complete external sensor data and high-definition maps. By focusing on key trajectory features (such as curvature, speed profiles, and maneuver type) already computed by the trajectory planner, the system achieves accurate situation awareness while eliminating the complexity of integrating and processing multiple external data sources.

The patent introduces the trajectory plan as an intermediary that mediates between the complex external environment perception (sensors and maps) and the manoeuvring controller configuration selection. Instead of directly processing sensor data and maps to determine controller settings, the system uses the pre-computed trajectory plan—which already encapsulates the driving situation—as the basis for configuration selection, thereby simplifying the system architecture while maintaining accurate situation awareness.

The patent applies local quality by selecting different controller configurations tailored to specific trajectory characteristics and driving situations. For example, it uses higher gain settings and more aggressive control parameters for precision manoeuvres requiring high accuracy (such as tight lane changes or parking), while switching to lower gain settings and smoother control profiles for comfort-critical situations (such as gentle lane changes or straight-line cruising). This localized optimization ensures that control accuracy and comfort are both maximized in their respective contexts.

The patent dynamically adjusts the controller configuration to balance accuracy and comfort in real-time based on the driving situation. The system transitions between different control strategies (such as switching between aggressive and conservative PID gains, or adjusting model predictive control weightings) according to the current trajectory characteristics, thereby achieving an optimal trade-off between control accuracy and occupant comfort that adapts to the immediate driving context rather than being fixed.