VertiAdaptor: Online Kinodynamics Adaptation for Vertically Challenging Terrain

2026-03-06

AI Summary

This work proposes an online adaptive framework to address the limited generalization of motion dynamics models for autonomous vehicles operating in complex, unstructured off-road environments characterized by highly variable terrain. The approach uniquely integrates elevation and semantic information into a terrain-aware dynamics model by embedding them within the dynamical space spanned by basis functions of a Neural Ordinary Differential Equation (Neural ODE). Fast adaptation is achieved through a functional encoder combined with online least-squares optimization. Evaluations on the Chrono-based Verti-Bench simulation platform and real-world quadrupedal robot experiments demonstrate that the method improves prediction accuracy by up to 23.9% on unseen terrains and accelerates adaptation speed by a factor of five compared to existing approaches.

Technology Category

• Intelligent Robots: State Estimation
• Search and Optimization: Sampling/Simulation-based Search
• Planning, Routing, and Scheduling: Model-Based Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Autonomous driving in off-road environments presents significant challenges due to the dynamic and unpredictable nature of unstructured terrain. Traditional kinodynamic models often struggle to generalize across diverse geometric and semantic terrain types, underscoring the need for real-time adaptation to ensure safe and reliable navigation. We propose VertiAdaptor (VA), a novel online adaptation framework that efficiently integrates elevation with semantic embeddings to enable terrain-aware kinodynamic modeling and planning via function encoders. VA learns a kinodynamic space spanned by a set of neural ordinary differential equation basis functions, capturing complex vehicle-terrain interactions across varied environments. After offline training, the proposed approach can rapidly adapt to new, unseen environments by identifying kinodynamics in the learned space through a computationally efficient least-squares calculation. We evaluate VA within the Verti-Bench simulator, built on the Chrono multi-physics engine, and validate its performance both in simulation and on a physical Verti-4-Wheeler platform. Our results demonstrate that VA improves prediction accuracy by up to 23.9% and achieves a 5X faster adaptation time, advancing the robustness and reliability of autonomous robots in complex and evolving off-road environments.

Problem

• off-road autonomous driving
• kinodynamic adaptation
• terrain generalization
• unstructured terrain
• real-time navigation

Innovation

• online adaptation
• kinodynamic modeling
• neural ODE
• terrain-aware planning
• semantic elevation fusion