Abstract Radar/Lidar and vision sensors have complementary properties. In this chapter, we consider multi-sensor multi-object detection and tracking systems to improve overall system performance. We formulate the probability framework of tracking and system model based on EKF using a single sensor. For multi-sensor data, data association is the first of all steps after receiving new data, aiming at judging the corresponding relation between the current observation and the previous track. We classify the data association into observation-to-observation and observation-to-track. Then we use Global Nearest Neighbor (GNN) algorithm to associate the observation with another observation or a track. For multi-sensor fusion, we map the lidar coordinates and radar coordinates into the vehicle coordinates to solve the position alignment and synchronize time between them by using the prediction equation of EKF. In comparison with other similar systems, our framework is more concise and efficient.
Multiple-Sensor Based Multiple-Object Tracking
2011-01-01
15 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Multiple Sensor Fusion and Classification for Moving Object Detection and Tracking
Online Contents | 2016
|Multiple sensor tracking architecture comparison
IEEE | 2014
|