Exploration of Parking Guidance based on Vehicle Crowdsourcing

The development of connected vehicles will bring new solution to the urban parking guidance systems. In this paper, we propose a new idea of guiding vehicles to proper carparks by the vehicles crowdsourcing. Meanwhile, a conditional LSTM is introduced to perform parking guidance from the perspective of classification, trained by the crowdsourced data and the information mined from them. The experiment shows the proposed method is not only practicable, but also adapt to the varying parking circumstance within the region, which would take advantage over than the parking guidance before. In addition, the method is crowdsourcing, which means its costs of deployment and maintenance are much less than collecting real-time information of parking spaces all over the city. In our opinion, this method is pretty competitive and promising in the era of connected vehicles.