Microwave-based vehicle detection products from two manufacturers were selected for field testing and evaluation: Wavetronix and Intersector. The two systems were installed by the manufacturer/distributor at a signalized intersection. Initial evaluation was performed and the results were shared with the companies. They were given an opportunity to change or fine-tune the systems setup, if they wanted, resulting in a modified setup. Results are presented in this report in terms four types of errors (false, missed, stuck-on, and dropped calls). At the stop bar, at least 94% of detections for Wavetronix and 96% for Intersector were correct. At stop bar zones, the overall occurrence of false calls for Wavetronix ranged from 0.56% to 1.62%. Missed calls were low for Zones 1 and 2 (0.13% and 0.43%) but significantly higher in Zone 3 (6.05%). Also, stuck-on calls were only observed in Zone 3 (0.58%), and a few dropped calls were found almost exclusively in Zone 3 (0.16%). For Intersector, false calls ranged from 1.4% to 3.56% and missed calls ranged between 0.05% and 0.27%. Stuckon calls ranged from 0.92% for 2.83% and dropped calls were very low (0% and 0.19%). At the advance zones, at least 91% of detections for Wavetronix and 99% for Intersector were correct. For the advance zone, a direct comparison of the two systems was not performed because Wavetronix covered all three lanes combined, but Intersector had one zone covering only the center lane. Wavetronix did not have any stuck-on or dropped calls, missed calls were 1.07%, and false calls were 8.29% for the summer and fall datasets combined. Intersector had no dropped calls, 0.04% stuck-on calls (only one call), 0.8% missed calls, and 0.7% false calls. Additional testing is under way to evaluate the performance of the two systems under inclement weather conditions. (Author/publisher)
Abstract