Real-time Driver Advisory Model: Intelligent Transportation Systems

A vehicle driver is an agent operating under partially observable, dynamic, nondeterministic and multiagent environments. This poses a challenge to drivers operating on unfamiliar roads, hence compromised road safety. It is therefore paramount to explore mechanisms for driver advisory on the status of roads ahead. This paper presents a model for providing real-time advisory alerts to drivers approaching mapped points of interest (POIs) and/or experiencing overspeeding behaviour. POIs include speed-limited zones, intersections, speed bumps and black spots. Determination of driver's approaching POIs used the K-Nearest Neighbour algorithm centered on the Spherical Law of Cosines. A text-to-speech Android app read text SMS alerts to avoid diversion of driver's attention. The model is kind of a Vehicular Ad-hoc Network as a low cost Vehicle-to-Driver communication using GPS, GSM and GIS. It is applicable to low-end infrastructure in developing nations. Experiments for model validation yielded positive results with success rates above 78% in terms of alert messages delivered to drivers at good distances for better reaction times. The study was carried out as a technical test of configurable technology that supports elements of Intelligent Transportation Systems, whose implementation will influence on driving behaviour, hence improving on performance and road safety.