New Robust Multiple Pedestrian Detection and Tracking System
Large Scale Field Tests of MigmaMidblockTM
Under FHWA SBIR Phase I/II funding, we have developed a product, MigmaMidblockTM, using stereoscopic technologies for the pedestrian detection at mid-block crossings and signalized intersections. It has been fielded at a mid-block crossing at Phillips Academy located in Andover, MA. This product can trigger the beacon flashers once pedestrians (i.e., high school students at Phillips Academy) are detected in the detection zone. After the field test over one month, Phillips Academy has decided to keep the system for its use and plans to purchase two more units to replace its existing IR and microwave detectors.
Before MigmaMidblockTM was installed, the same mid-block crossing has an IR-based detector installed. School has also installed the IR-based detector and microwave detector at two other mid-block crossings. It has been observed that there are many issues related to IR and microwave detectors, including missed detections and frequent false calls resulted from pedestrians walking on the sidewalk. Phillips Academy has confirmed that our product outperforms both IR and microwave detectors.
The performance issues related to IR and microwave detectors have been reported by numerous studies. IR and microwave technologies are currently recognized as two best candidates for outdoor pedestrian detections. Although stereo vision based pedestrian detection technologies have been studied over the past decades, there are currently no field-proven commercial stereoscopic products for pedestrian detections. Our product represents the cutting edge technology for pedestrian detection in the outdoor environment.
Although our limited field test results have shown that our product has a great potential, extensive field tests under various weather conditions are needed for it to be commercially viable. Various hardware and software issues unforeseen during the product design can be exposed during the field tests and can be resolved in a timely fashion. Performance evaluation and benefit analysis, especially for visually impaired pedestrians, can also help ease the concerns of the effectiveness of this product. Moreover, results from large scale field tests can help future research in the field of pedestrian detections.
* information listed above is at the time of submission.