Enhancing in vehicle digital maps via GPS crowdsourcing

In this paper, we propose a simple and effective method to extend digital maps with the location and timing of stop-signs and traffic lights in a city, given GPS traces collected by on-road vehicles. Our system finds the location and timing of traffic lights and stop-signs using a small number of traces per road segment. We developed and evaluated the system by applying it to two sets of GPS traces - open street map, a multicity, publicly available database of GPS traces, and a set of traces collected in west Los Angeles. Evaluation results show that our system can estimate the location of stop-signs with as little as 5 traversal per road segment and the location/timing of traffic lights with as little as 7 traversals, with more than 90% accuracy.