How many probe vehicles do we need to collect on-street parking information?
Authors:
- 1 June 2017
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 538-543
Abstract
Finding a parking space is a key mobility problem in urban scenarios. Intelligent Transportation Systems with up-to-date information about the state of parking infrastructure could significantly mitigate this problem, by guiding drivers towards locations with high likelihood of free spaces. The challenging part is how to obtain accurate, near real-time, information on the availability of on-street parking spaces. An encouraging solution is to crowd-sense such availability with a fleet of probe vehicles, like taxis, exploiting on-board sensors as side-scanning ultrasonic sensors or windshield-mounted cameras. Anyhow, while the spatio-temporal patterns of taxis are well studied, the required number of taxis for crowd-sensing parking availability has never been deeply investigated. In this paper we present an empirical evaluation of the suitability of a fleet of taxis as probes for crowd-sensing. In particular, we combined a dataset of trajectories collected from 536 taxis over three weeks, with one of parking availability collected from 8,000 sensors embedded in the asphalt, both from San Francisco, USA. By using these data, we provide an experimental evaluation of the quality of on-street parking information that could be achieved by fleets of taxis of different sizes. Results show that a fleet of 300 vehicles is enough to cover a urban scenario like San Francisco downtown, showing an error against the infrastructure sensors bigger than ±1 parking spaces, only in less than 15% of the cases.Keywords
This publication has 12 references indexed in Scilit:
- A Survey of Smart Parking SolutionsIEEE Transactions on Intelligent Transportation Systems, 2017
- Spatio-Temporal Road Coverage of Probe Vehicles: A Case Study on Crowd-Sensing of Parking Availability with Taxis2017
- What are the potentialities of crowdsourcing for dynamic maps of on-street parking spaces?2016
- An Architecture for a Mobility Recommender System in Smart CitiesProcedia Computer Science, 2016
- UPDetector2014
- Temporal and Spatial Clustering for a Parking Prediction Service2014
- On-vehicle video-based parking lot recognition with fisheye optics2013
- Real-Time Street Parking Availability Estimation2013
- ParkNet2010
- Map-matching for low-sampling-rate GPS trajectories2009