基于视锥距离和自适应权重卡尔曼滤波的 多传感器融合算法

Translated title of the contribution: A Multi-sensor Fusion Algorithm Based on View-cone Distance and Adaptive WeightedKalman Filter

Jie Li*, luo Wei Zhang, Xiao Yan Wang, Zheng Hu, Hai Lan, Zhi Yong Wang, Li Wang

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

Abstract

To improve the environmental perception ability of intelligent driving systems, multimodal sensors have been integrated with the artificial intelligence technology to solvethe problem of single-modal sensors, such as poor recognition and vulnerability to interference in environmental perception. However, problems of feature matchingbetween cross-modal sensors, such as inconsistent feature representations, sensing errors, and delay errors, persist. To address these problems, this study proposes aview-cone distance metric-based method, constructs a target-matching matrix, and uses the Hungarian algorithm for inter-frame association matching. Based on theintersection over union (IOU) and Euclidean distance metrics, the multiple object tracking accuracy (MOTA) of the proposed method improved to 81.22% and 58.62% inthe cyclist and pedestrian categories, respectively. An adaptive weight-adjustment technique was also introduced to optimize the Kalman filter algorithm so that low-complexity and efficient cross-modal sensor fusion target detection and tracking could be achieved. Compared with the individual camera and light detection and ranging(LIDAR) predictions, the root mean square error of the fusion method reaches 0.3490, denoting a reduction by 30.37% and 30.53% compared with the camera and LIDARmethods, respectively, and confirming the accuracy of the proposed adaptive weighting Kalman filter fusion tracking method. The multi-target tracking experimental testconducted on the KITTI dataset achieved an accuracy of 88.25%, comparable to the performance of current mainstream methods. The test results under multiple weatherconditions also demonstrated excellent performance, with target detection accuracies of 96.40%, 75.51%, and 91.87% for vehicles, pedestrians, and cyclists, respectively.Compared to a single sensor, the fusion method attained superior detection results under multiple road conditions, improved the reliability and robustness of the system,and laid a solid foundation for the further development of driverless technology.

Translated title of the contributionA Multi-sensor Fusion Algorithm Based on View-cone Distance and Adaptive WeightedKalman Filter
Original languageChinese (Traditional)
Pages (from-to)194-203
Number of pages10
JournalZhongguo Gonglu Xuebao/China Journal of Highway and Transport
Volume37
Issue number3
DOIs
Publication statusPublished - 31 Mar 2024
Externally publishedYes

Fingerprint

Dive into the research topics of 'A Multi-sensor Fusion Algorithm Based on View-cone Distance and Adaptive WeightedKalman Filter'. Together they form a unique fingerprint.

Cite this