Model Validation of a Low-Speed and Reverse Driving Articulated Vehicle

TL;DR

This paper evaluates different vehicle models for accurately predicting the behavior of articulated vehicles during low-speed and reverse maneuvers, crucial for autonomous operations at distribution centers.

Contribution

It introduces a stabilization approach for unstable system poles and compares three vehicle models against real-world data for reverse driving validation.

Findings

Non-linear single track model outperforms others at large articulation angles

Open-loop validation is unsuitable for reverse maneuvers due to instability

The proposed approach improves model accuracy for autonomous vehicle control

Abstract

For the autonomous operation of articulated vehicles at distribution centers, accurate positioning of the vehicle is of the utmost importance. Automation of these vehicle poses several challenges, e.g. large swept path, asymmetric steering response, large slide slip angles of non-steered trailer axles and trailer instability while reversing. Therefore, a validated vehicle model is required that accurately and efficiently predicts the states of the vehicle. Unlike forward driving, open-loop validation methods can not be used for reverse driving of articulated vehicles due to their unstable dynamics. This paper proposes an approach to stabilize the unstable pole of the system and compares three vehicle models (kinematic, non-linear single track and multibody dynamics model) against real-world test data obtained from low-speed experiments at a distribution center. It is concluded that single track non-linear model has a better performance in comparison to other models for large articulation angles and reverse driving maneuvers.