AI Model Predicts Parking Lot Crowding With New Accuracy
Researchers have developed a forecasting system that predicts when parking lots will fill up by analyzing how drivers actually arrive and leave. The advance could help cities reduce traffic congestion and emissions by directing drivers to available spaces faster, while cutting the computational costs of existing prediction methods.
Originaltitel: New Forecasting Framework for Occupancy Dynamics of Car Parking Lots by Integrating Arrival and Departure Behaviors
This paper introduces a novel framework for predicting parking lot occupancy dynamics in urban environments, and offers insights for the planning of sustainable urban mobility. By stochastically modeling the parking lot dynamics, together with the integration of drivers’ behavioral changes in arrival and departure patterns over time, the proposed framework offers a comprehensive solution to address the challenges associated with forecasting the parking occupancy. Key features of our framework include the application of queueing theory for precise occupancy distribution, a Markov regime switching model to capture arrival and departure dynamics respectively, and interpretability with minimal unknown parameters. Extensive experiments with real-world datasets are conducted to validate the framework’s accuracy and interpretability, showcasing its superiority in forecasting accuracy and computational efficiency over existing methods.