A novel algorithm to track closely spaced road vehicles using a low density flash lidar

This paper develops a novel algorithm for tracking closely-spaced road vehicles using a low-density flash lidar. Low-density flash lidars are recent to the automotive market and have attracted attention due to their low cost. However, these sensors have a poor angular resolution which makes tracking the lateral motion of targets challenging. One such challenge, namely unresolved measurements from multiple vehicles, is addressed in this paper. Detections from multiple targets can become unresolved if the targets are closely spaced. Traditional tracking algorithms, which do not account for unresolved measurements, lead to degraded tracking performance in such scenarios. This paper proposes a novel method based on predicting the lateral motion states of targets forward in time, and truncating their probability density functions at points based on the unresolved measurements. The method proposed works with a computationally simple data association algorithm, requires no sensor modeling, and can track more than two closely spaced targets. The proposed algorithm is evaluated using simulations and on-road experiments, and the results demonstrate its capability in maintaining target tracks even in the presence of unresolved detections. Further, a comparison with the well known Joint Probabilistic Data Association-Merged algorithm showed that the proposed algorithm is computationally lighter. (c) 2021 Elsevier B.V. All rights reserved.

Automated summary

This paper presents a novel algorithm for tracking closely-spaced road vehicles using low-density flash lidar. The algorithm addresses the challenge of unresolved measurements from multiple targets, by predicting target states and truncating probability density functions based on unresolved measurements. It demonstrates the capability to track multiple closely spaced targets effectively.