Mixed kinematic and dynamic sideslip angle observer for accurate control of fast off‐road mobile robots

Lenain, Roland; Thuilot, Benoit; Cariou, Christophe; Martinet, Philippe · 2009 · OpenAlex-citations

DOI: 10.1002/rob.20319

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the challenge of maintaining high-precision path tracking for fast off-road mobile robots operating on low-grip terrain. While previous adaptive and predictive control algorithms based on extended kinematic models achieved accuracy within ±10 cm at speeds below 3 m/s, they suffered from significant estimation delays and oscillations at higher velocities (up to 10 m/s). These performance degradations were attributed to the inability of purely kinematic observers to reactively estimate sideslip angles during transient phases, such as curvature transitions. The authors propose a mixed kinematic and dynamic observer to integrate dynamic effects into sideslip estimation, thereby reducing latency without requiring difficult-to-measure dynamic vehicle parameters. The proposed method employs a backstepping approach combining kinematic and dynamic models. First, a preliminary extended kinematic observer estimates slow-varying sideslip angles, providing a steady-state approximation. These estimates are injected into a dynamic observer based on a linearized bicycle model, which accounts for yaw rate and vehicle sideslip. Crucially, the dynamic model incorporates on-line adaptation of front and rear cornering stiffnesses to reflect varying grip conditions and non-linear tire behavior. This adaptation is achieved by tuning observer gains to prioritize the convergence of observed yaw rate with measured values, allowing the stiffness parameters to reactively describe terrain interactions. The resulting dynamic observer supplies reactive and reliable sliding variable estimates. The algorithm was evaluated through advanced simulations coupling Adams and Matlab software to investigate high-speed capabilities, as well as actual experiments on a mobile robot platform. Simulations demonstrated that the mixed observer significantly reduces the delay in sideslip angle estimation compared to the purely kinematic approach. Experimental results, conducted at the platform’s maximum velocity, confirmed the benefits of the proposed approach, showing improved stability and reduced oscillations during path tracking. The method preserves high accuracy even when grip conditions change rapidly, addressing the limitations of previous kinematic-only observers. The significance of this work lies in enabling accurate control of autonomous mobile robots at higher speeds in unstructured environments. By integrating dynamic effects into the observation process while avoiding the need for precise, static dynamic parameters, the approach bridges the gap between simple kinematic control and complex dynamic modeling. This allows for robust path tracking in applications such as agriculture, surveillance, and military operations, where vehicles must operate efficiently on variable terrain at speeds previously deemed too high for precise kinematic control. The study demonstrates that reactive estimation of sliding variables is critical for maintaining control accuracy as vehicle velocity increases.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• steering pattern