Multi-Object Tracking Strategy of Autonomous Vehicle Using Modified Unscented Kalman Filter and Reference Point Switching

Wang, Muyuan; Xiaodong, WU

doi:10.1007/s12204-021-2350-0

Cited by 3 publications

(2 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The limitation of this work is that they use a naive KF model that assumes objects always have a constant velocity, which does not apply to objects like cars. Wang [18] and Kim [19] apply non-linear filters to estimate complex motion for tracked objects. Kim [19] uses object distance from LiDAR and Radar sensors to track objects utilizing the extended KF and shows the result by operating Prescan simulator [24] in different scenarios.…”

Section: Literature Reviewmentioning

confidence: 99%

“…Kim [19] uses object distance from LiDAR and Radar sensors to track objects utilizing the extended KF and shows the result by operating Prescan simulator [24] in different scenarios. In contrast, Wang [18] proposes a modified version of the unscented KF for state estimation that improves tracking accuracy. Meanwhile, Weng et al [25] use baseline algorithms, Hungarian algorithm [13], and show the capability of baseline algorithms to achieve a comparable tracking accuracy to the proposed deep learning solutions.…”

Section: Literature Reviewmentioning

confidence: 99%

See 1 more Smart Citation

DFR-FastMOT: Detection Failure Resistant Tracker for Fast Multi-Object Tracking Based on Sensor Fusion

Nagy¹,

Khonji²,

Dias³

et al. 2023

Preprint

View full text Add to dashboard Cite

Persistent multi-object tracking (MOT) allows autonomous vehicles to navigate safely in highly dynamic environments. One of the well-known challenges in MOT is object occlusion when an object becomes unobservant for subsequent frames. The current MOT methods store objects information, like objects' trajectory, in internal memory to recover the objects after occlusions. However, they retain shortterm memory to save computational time and avoid slowing down the MOT method. As a result, they lose track of objects in some occlusion scenarios, particularly long ones. In this paper, we propose DFR-FastMOT, a light MOT method that uses data from a camera and LiDAR sensors and relies on an algebraic formulation for object association and fusion. The formulation boosts the computational time and permits long-term memory that tackles more occlusion scenarios. Our method shows outstanding tracking performance over recent learning and non-learning benchmarks with about 3% and 4% margin in M OT A, respectively. Also, we conduct extensive experiments that simulate occlusion phenomena by employing detectors with various distortion levels. The proposed solution enables superior performance under various distortion levels in detection over current state-of-art methods. Our framework processes about 7, 763 frames in 1.48 seconds, which is seven times faster than recent benchmarks. The framework will be available at https://github.com/MohamedNagyMostafa/DFR-FastMOT.

show abstract