Authors :
Himesh Chauhan; Varun Choudhary; Syed Faizan Haider; Dr. Gokulnath C
Volume/Issue :
Volume 10 - 2025, Issue 5 - May
Google Scholar :
https://tinyurl.com/3jxn4vek
DOI :
https://doi.org/10.38124/ijisrt/25may495
Google Scholar
Note : A published paper may take 4-5 working days from the publication date to appear in PlumX Metrics, Semantic Scholar, and ResearchGate.
Note : Google Scholar may take 15 to 20 days to display the article.
Abstract :
Autonomous vehicles (AVs) need complex perception systems for safe operation under dynamic traffic scenes. We
introduce TrajecTrack, a machine learning-based platform that integrates real-time trajectory estimation, velocity
estimation and lane detection from LiDAR and vision inputs. We apply DBSCAN clustering and the constant velocity model
for predicted trajectories, with our speed estimation based on YOLOv8 and ByteTrack, plus a new module for lane detection
based on edge detection and the Hough transform. Compared to the NuScenes dataset and sample video input, TrajecTrack
provides high-accuracy visualizations of the trajectories, velocities and road lane markings and therefore improves the
situational awareness of AVs. This paper contributes significantly to the field of AV perception in that it supports a scalable
single solution, with future implications being in traffic violative detection.
Keywords :
Trajectory Estimation, Speed Estimation, Lane Detection, Autonomous Driving, LiDAR, Camera-Based Perception, DBSCAN, YOLOv8, ByteTrack, Hough Transform, NuScenes Dataset, Real-Time Analysis.
References :
- Caesar, H., et al., “nuScenes: A Multimodal Dataset for Autonomous Driving,” CVPR, 2020.
- Redmon, J., et al., “YOLOv3: An Incremental Improvement,” arXiv:1804.02767, 2018.
- Zhang, Y., et al., “ByteTrack: Multi-Object Tracking by Associating Every Detection Box,” arXiv:2110.06864, 2021.
- Ester, M., et al., “DBSCAN: A Density-Based Algorithm,” KDD, 1996.
- Kuhn, H. W., “The Hungarian Method for the Assignment Problem,” Naval Research Logistics, 1955.
Autonomous vehicles (AVs) need complex perception systems for safe operation under dynamic traffic scenes. We
introduce TrajecTrack, a machine learning-based platform that integrates real-time trajectory estimation, velocity
estimation and lane detection from LiDAR and vision inputs. We apply DBSCAN clustering and the constant velocity model
for predicted trajectories, with our speed estimation based on YOLOv8 and ByteTrack, plus a new module for lane detection
based on edge detection and the Hough transform. Compared to the NuScenes dataset and sample video input, TrajecTrack
provides high-accuracy visualizations of the trajectories, velocities and road lane markings and therefore improves the
situational awareness of AVs. This paper contributes significantly to the field of AV perception in that it supports a scalable
single solution, with future implications being in traffic violative detection.
Keywords :
Trajectory Estimation, Speed Estimation, Lane Detection, Autonomous Driving, LiDAR, Camera-Based Perception, DBSCAN, YOLOv8, ByteTrack, Hough Transform, NuScenes Dataset, Real-Time Analysis.
