Details of Research Outputs

Vehicle detection in autonomous driving could be very challenging under adverse road conditions. The problem has been studied intensively. However, recent studies have shown that the problem remains unsolved, especially when the vehicles are occluded or under low-light conditions. This paper adopts a different approach to vehicle detection by taking advantage of RFID technology. Specifically, RFID tags are attached to the vehicle's surfaces, and then a system is designed to detect, locate, and track those tags dynamically. In addition, RFIDs are allowed to store user data on chips. To fully utilize this feature, this paper develops an algorithm to select and store the most critical information in tags for recovering the boundaries of occluded vehicles and finding the vehicle’s location and orientation. The proposed method achieves the following objectives: (1) Vehicles could be detected at a relatively long distance in any conditions (including low-light or adverse weather). (2) The boundary of the occluded vehicle could be recovered. (3) Vehicles are still detectable even if they are turned off. (4) The implementation is relatively simple. The evaluation results have shown that the proposed method is able to detect a vehicle's orientation and rotation and recover the boundary for an occluded vehicle.