Detection Performance of Wireless Magnetometers at Signalized Intersection and Railroad Grade Crossing under various Weather Conditions

The performance of wireless magnetometers for vehicle detection was evaluated under favorable and adverse weather conditions at two sites: (a) a signalized intersection with stop bar and advance detection zones and (b) a railroad grade crossing, as a potential backup system to control the operation of gates. Loop detectors and magnetometers were installed at the same locations, providing similar detection areas. Discrepancies between activations from loops and magnetometers provided pointers to potential errors in the systems, and video images were used to verify them visually. At the signalized intersection, the most common type of detection error by the magnetometers in the stop bar zones was a false call caused by vehicles in the adjacent lane, with 5.6% to 7.6% error per lane in favorable weather and 7.7% to 15.4% in winter conditions. At the advance zones the most frequent error in all weather conditions was a missed call caused by vehicles traveling between two lanes, ranging from 2.9% to 9.7% in the left-turn lanes. At the railroad grade crossing, however, most errors were false and stuck-on calls. On average, there was one stuck-on call per every 150 trains, and one per every 2,800 vehicles. False calls varied from 13.4% to 14.1% in the through lane and from 55.9% to 59.9% in the left-turn lane, caused mostly by vehicles traveling in the opposing direction (this was an atypical location with no median separating the two directions of traffic).