Stair Climbing for Vehicles with Articulated Tracked Arms: Closed-Loop Flippers Control
Thales Henriques da Silva, Fernando Lizarralde
AI summary
Problem
Current stair-climbing algorithms for hybrid tracked robots rely on heuristic flipper positioning or require precise contact-point data, making autonomous traversal complex and prone to collisions.
Approach
The authors model the articulated flippers as a constrained kinematic chain and design a state-feedback control law that regulates flipper heights based solely on robot orientation and detected stair pitch, coordinated via a finite state machine.
Key results
- Derivation of a differential kinematic model for flippers constrained to stair planes
- Design of a globally exponentially stable state-feedback control law for autonomous flipper height regulation
- Implementation of a finite state machine to manage height calculations during floor transitions
- Experimental validation on the AATV Rosi robot demonstrating smooth stair ascent with zero steady-state height error and successful collision avoidance
Why it matters
Enables safer, fully autonomous navigation for hybrid tracked robots in unstructured environments without complex terrain mapping or teleoperation.
Abstract
Stair-climbing algorithms for tracked robots often neglect proper control of articulated arms, relying instead on a set of empirically predefined positions. In contrast, this work introduces a novel closed-loop control approach for mobile robots equipped with actively articulated tracked arms during stair climbing. The robot under consideration employs either wheels or tracks as its primary locomotion system, while the arms can be actuated to extend mobility when required. To perceive the environment, the robot is equipped with a depth sensor for stair detection and an IMU used exclusively for orientation estimation. The configuration of each lateral arm set is planned by modeling its differential kinematics as a planar kinematic chain subject to position constraints. Based on this model, a state-feedback control law is designed to guarantee stability and convergence, allowing the robot to climb stairs autonomously. The controller adapts the tracks to the stair ge- ometry, enabling maneuvers that resemble a snake-like climbing motion. This improves traction, prevents abrupt movements, and avoids undesired floor collisions. The effectiveness of the proposed control scheme is demonstrated on a real robot, with experimental results validating its performance.