Completed

Object Tracking via Multimodal Sensor System - Tri-Kalman Approach

Robust tracking using tri-Kalman filters and decision-level sensor fusion

Overview

With the recent hike in the autonomous and automotive industries, sensor-fusion based perception has garnered significant attention for multi-object classification and tracking applications. This project presents a robust tracking framework using high-level monocular-camera and millimeter wave radar sensor-fusion. The method aims to improve localization accuracy by leveraging radar's depth and camera's cross-range resolutions using decision-level sensor fusion, and make the system robust by continuously tracking objects despite single sensor failures using a tri-Kalman filter setup. The approach offers promising MOTA and MOTP metrics with significantly low missed detection rates.

Key Highlights

Sensor Fusion
Robust Tracking
Tri-Kalman Filter
Decision-Level Fusion
Low Missed Detection

Methodology

Decision-level sensor fusion using tri-Kalman filter framework. Camera intrinsic calibration for bird's-eye view estimation. Hungarian Algorithm for measurement association.