A Global Route Planning Framework for AVP Based on V2X Communication with Timely Effective Information

Automated Valet Parking (AVP) is a feasible L4 technology in autonomous parking lot scenarios benefiting from V2I technology. Existing global route planning methods in AVP tend to rely on high-level intelligent parking lots' infrastructure as a single source of information to obtain real-time berth information. However, the intelligence level of most parking lots is insufficient, whose communication spatial range and real-time performance is limited, making these methods fail to obtain essential information to plan. To address the issues of information acquisition and real-time requirements, we propose a timely effective route planning framework based on V2X(TERP-V2X). It utilizes V2I, V2V, and self-perception to jointly obtain parking information and predict future berth state by current information considering the timely effectiveness of V2X information. TERP-V2X combines road cost in form of probability model with time-varying parameters and orthogonal linked list to establish an information model. We use Bayesian method to update the information model and use breadth-first global route planning method with random and uncertain road cost to find free berth. By TERP-V2X, we expand information acquisition sources and weaken the rely on real-time information, suitable for parking lots of various levels of intelligence. The experiment results conducted in VISSIM show that TERP-V2X effectively plans the parking route and can be applied in various V2X penetration rate of AVP participants compatibly with less time consumed on travelling to find berth.