LADAR-Based Vehicle Tracking and Trajectory Estimation for Urban Driving

TL;DR

This paper presents a LADAR-based system for accurate vehicle detection and trajectory estimation in urban environments, enabling safe autonomous navigation through multi-model tracking and real-time processing.

Contribution

It introduces a novel multi-hypothesis tracking approach combining variable-axis Ackerman and independent steering models for improved trajectory estimation.

Findings

Achieved real-time performance on a moving vehicle.

Demonstrated reliable vehicle tracking in urban traffic.

Enabled autonomous vehicle navigation using estimated trajectories.

Abstract

Safe mobility for unmanned ground vehicles requires reliable detection of other vehicles, along with precise estimates of their locations and trajectories. Here we describe the algorithms and system we have developed for accurate trajectory estimation of nearby vehicles using an onboard scanning LADAR. We introduce a variable-axis Ackerman steering model and compare this to an independent steering model. Then for robust tracking we propose a multi-hypothesis tracker that combines these kinematic models to leverage the strengths of each. When trajectories estimated with our techniques are input into a planner, they enable an unmanned vehicle to negotiate traffic in urban environments. Results have been evaluated running in real time on a moving vehicle with a scanning LADAR.