Multi-target tracking method and system based on multiple sensors and computer readable medium

Abstract

The invention discloses a multi-target tracking method based on multiple sensors. The multi-target tracking method comprises the following steps: updating and predicting the state of a tracked target;detecting target association and updating a tracked target set; updating and predicting the state of the tracked target, and updating the state value of the tracked target through the determined current sensor feedback state value of the tracked target; predicting a state value of the tracked target when the sensor feeds back the next time to obtain a state prediction value of the tracked target,the detection target association is association of a detection target and a tracked target, and the detection target matched with the tracked target is screened out from the detection targets; and updating the tracked targets in the tracked target set according to the final result of the detection target association. The invention also provides a multi-target tracking system based on multiple sensors. According to the invention, obstacles around the main vehicle can be accurately tracked, the possibility of false alarm and missing detection is greatly reduced, and the effectiveness of multi-target tracking is ensured, and the robustness is high.

Description

technical field [0001] The invention belongs to the field of automatic driving, and in particular relates to a multi-sensor-based multi-target tracking method, system and computer-readable medium. Background technique [0002] Environmental awareness is a key technology in the field of autonomous driving, which plays a very important role in the safety of autonomous driving. At present, sensors such as Lidar, millimeter-wave radar Radar, and camera are commonly used in the field of autonomous driving to detect obstacles such as pedestrians and vehicles around the vehicle. The data detected by the sensor is often processed by deep neural network or traditional feature extraction methods; however, the commonly used methods cannot guarantee that each frame can give correct detection results, and false alarms and missed detections often occur. At the same time, millimeter-wave radar Radar is easily affected by surrounding metal objects, and its detection output is unstable. In...

AI Technical Summary

Problems solved by technology

The data detected by the sensor is often processed by deep neural network or traditional feature extraction methods; however, the commonly used methods cannot guarantee that each frame can give correct detection results, and false alarms and missed detections often occur.

At the same time, millimeter-wave radar Radar is easily affected by surrounding metal objects, and its detection output is unstable

In addition, in the tracking algorithm, the end-to-end multi-target tracking neural network has high requirements on the computing platform, and it is difficult to implement a large amount of sample data at the same time, and it cannot guarantee 100% tracking effect

There is no way to accurately track the obstacles around the vehicle will be a big safety hazard for the vehicle's automatic driving

Images