An Ego-Based Approach to Planning and Control for Automated Valet Parking Applications

This paper introduces an ego-based approach to automated valet parking in low-complexity scenarios. The proposed solution aims at realizing the valet parking application by exploiting a minimum amount of information provided by the infrastructure and implementing all the required driving functions based on proprioceptive sensor data. An encapsulated hierarchical architecture is introduced to accomplish this aim. The higher hierarchical level, i.e. the Global Planner, computes a feasible and robust geometric path from the drop-off area to the parking destination. At the lower level, the Local Planner based on Model Predictive Control and Artificial Potential fields, tracks the path and realizes the final parking maneuver. Decision-making during vehicle maneuvering is implemented by a suitable behavioral logic that, based on sensor-acquired data, manages vehicle interaction in specific situations such as, e.g., precedence in road intersections, and traffic jam handling. Extensive simulation results performed in realistic driving scenarios are introduced to show the effectiveness of the proposed approach.