Journal
IEEE ACCESS
Volume 7, Issue -, Pages 65944-65954Publisher
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/ACCESS.2019.2916718
Keywords
Deer crossing; object classification; roadside LiDAR; vehicle trajectories
Categories
Funding
- Natural Science Foundation of China [61463026, 61463027]
- Nevada Department of Transportation (NDOT) [P224-14-803/TO #13]
- Fundamental Research Funds for the Central Universities [300102238614]
- Ministry of Education of Humanities and Social Science Project [18YJAZH120]
Ask authors/readers for more resources
Deer crossing roads are a major concern of highway safety in rural and suburban areas in the United States. This paper provided an innovative approach to detecting deer crossing at highways using 3D light detection and ranging (LiDAR) technology. The developed LiDAR data processing procedure includes background filtering, object clustering, and object classification. An automatic background filtering method based on the point distribution was applied to exclude background but keep the deer (and road users if they exist) in the space. A modified density-based spatial clustering of applications with noise (DBSCAN) algorithm was used for object clustering. Adaptive searching parameters were applied in the vertical and horizontal directions to cluster the points. The cluster groups were further classified into three groups-deer, pedestrians, and vehicles, using three different algorithms: naive Bayes, random forest, and k-nearest neighbor. The testing results showed that the random forest (RF) can provide the highest accuracy for classification among the three algorithms. The results of the field test showed that the developed method can detect the deer with an average distance of 30 m far away from the LiDAR. The time delay is about 0.2 s in this test. The deer crossing information can warn drivers about the risks of deer-vehicle crashes in real time.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available