CAHSOR: Competence-Aware High-Speed Off-Road Ground Navigation in $mathbb {SE}(3)$

2024-02-10

IEEE Robotics and Automation Letters

AI Summary

In high-speed off-road navigation through unstructured terrain, the conventional SE(2) motion assumption fails, leading to vehicle instability—including rollovers, lateral skidding, and severe vibrations. Method: This paper proposes an SE(3)-aware autonomous navigation framework that explicitly models vehicle capability constraints in full 3D space. We develop a 6-DOF forward kinematic-dynamic model fused with multimodal sensory inputs (vision, wheel odometry, and IMU), and introduce TRON—a self-supervised terrain-vehicle interaction model enabling joint kinematic and dynamic prediction. Further, we integrate SE(3) Lie-group optimization with real-time terrain representation to jointly infer terrain adaptability and stability boundaries. Contribution/Results: To our knowledge, this is the first work to explicitly encode SE(3)-constrained vehicle capabilities in off-road navigation. Real-world experiments demonstrate a 62% reduction in instability events, with only an 8.6% decrease in average speed, while supporting both fully autonomous and human–machine shared control modes.

Technology Category

• Intelligent Robots: Localization, Mapping, and Navigation
• Search and Optimization: Sampling/Simulation-based Search
• Planning, Routing, and Scheduling: Model-Based Reasoning

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• 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

Abstract

While the workspace of traditional ground vehicles is usually assumed to be in a 2D plane, i.e., <inline-formula><tex-math notation="LaTeX">$mathbb {SE}(2)$</tex-math></inline-formula>, such an assumption may not hold when they drive at high speeds on unstructured off-road terrain: High-speed sharp turns on high-friction surfaces may lead to vehicle rollover; Turning aggressively on loose gravel or grass may violate the non-holonomic constraint and cause significant lateral sliding; Driving quickly on rugged terrain will produce extensive vibration along the vertical axis. Therefore, most offroad vehicles are currently limited to driving only at low speeds to assure vehicle stability and safety. In this letter, we aim at empowering high-speed off-road vehicles with competence awareness in <inline-formula><tex-math notation="LaTeX">$mathbb {SE}(3)$</tex-math></inline-formula> so that they can reason about the consequences of taking aggressive maneuvers on different terrain with a 6-DoF forward kinodynamic model. The kinodynamic model is learned from visual, speed, and inertial Terrain Representation for Off-road Navigation (<sc>tron</sc>) using multimodal, self-supervised vehicle-terrain interactions. We demonstrate the efficacy of our Competence-Aware High-Speed Off-Road (<sc>cahsor</sc>) navigation approach on a physical ground robot in both autonomous navigation and a human shared-control setup and show that <sc>cahsor</sc> can efficiently reduce vehicle instability by 62% while only compromising 8.6% average speed with the help of <sc>tron</sc>.

Problem

• Enabling high-speed off-road navigation in SE(3) for ground vehicles
• Addressing vehicle instability and safety on unstructured terrain
• Reducing rollover and sliding risks during aggressive maneuvers

Innovation

• 6-DoF kinodynamic model for SE(3) navigation
• Self-supervised terrain interaction learning
• TRON for visual-inertial terrain representation