Stand, Walk, Navigate: Recovery-Aware Visual Navigation on a Low-Cost Wheeled Quadruped
Jans Solano, Diego Quiroz
TL;DR
This work addresses autonomous navigation for a low-cost wheeled–quadruped by integrating a DRL-based locomotion policy with a fall-recovery module and lightweight vision-based perception. The approach uses proprioceptive sensing, a privileged terrain signal for learning, and NAV2 with 2D/3D costmaps to achieve goal-directed indoor navigation while maintaining robustness to disturbances. Key contributions include a recovery-aware training regimen, an integrated perception–navigation–locomotion stack, and extensive simulation results showing high recovery success and 98% indoor navigation success on budget hardware. The findings suggest that reliable autonomous operation is achievable on affordable platforms without premium actuators or sensors, enabling broader deployment and rapid testing of robust locomotion and navigation policies.
Abstract
Wheeled-legged robots combine the efficiency of wheels with the obstacle negotiation of legs, yet many state-of-the-art systems rely on costly actuators and sensors, and fall-recovery is seldom integrated, especially for wheeled-legged morphologies. This work presents a recovery-aware visual-inertial navigation system on a low-cost wheeled quadruped. The proposed system leverages vision-based perception from a depth camera and deep reinforcement learning policies for robust locomotion and autonomous recovery from falls across diverse terrains. Simulation experiments show agile mobility with low-torque actuators over irregular terrain and reliably recover from external perturbations and self-induced failures. We further show goal directed navigation in structured indoor spaces with low-cost perception. Overall, this approach lowers the barrier to deploying autonomous navigation and robust locomotion policies in budget-constrained robotic platforms.