Self-driving vehicles have attracted tremendous attention from all walks of life and parking prediction is essential for the vehicle to find a parking space. Many existing works focus on parking predictions in an sensor network equipped environment using large datasets. In this paper, we consider a campus-wide smart parking problem that requires no sensor installation and external large datasets. We assume that the parking availabilities of the garages are not known and the only information that a vehicle uses is its past recorded statistics. We get insights from the multi-armed bandits problem in reinforcement learning and come up with a framework for a vehicle to make a wise decision in each trial. We also propose several non-reinforcement learning algorithms for comparison. We conduct extensive simulation and discussion to evaluate the performance of these algorithms.
Energy-efficient Smart Parking for Self-driving Vehicles
2021-05-01
181748 byte
Conference paper
Electronic Resource
English
Smart parking system for vehicles
IEEE | 2014
|Parking management architecture for parking autonomous driving vehicles
European Patent Office | 2020
|