Authors: Riccardo Pieroni, Simone Specchia, Matteo Corno, Sergio Matteo Savaresi

Published on: March 06, 2024

Impact Score: 7.4

Arxiv code: Arxiv:2403.04112

Summary

• What is new: The local-based, map-agnostic multi-modal MOT approach combining camera and LiDAR data without requiring global pose information.
• Why this is important: Improving Multi-Object Tracking for self-driving cars by effectively incorporating both camera and LiDAR data.
• What the research proposes: A novel algorithm that uses a 3D object detector for camera data, classical clustering for LiDAR data, a three-step association process, and an Extended Kalman filter for motion estimation.
• Results: The algorithm demonstrated satisfactory results in both simulation and real-world data tests.

Technical Details

Technological frameworks used: Three-step association process, Extended Kalman filter

Models used: 3D object detector

Data used: Camera frames, LiDAR observations

Potential Impact

Automotive industry, particularly companies working on self-driving technology, and sensor manufacturers.

Want to implement this idea in a business?

We have generated a startup concept here: DriveAI.