Gait-Generation-Based Path Following Control of Snake Robots for Motion Recovery With Joint Locking Failure

Abstract

Snake robots operating in complex environments are susceptible to joint failures due to sustained high actuator loads, where large-angle joint locking severely distorts the gait shape and degrades motion performance. This paper presents an optimal path following control framework with gait generation for motion recovery of snake robots under locked-joint failures. Instead of relying on predefined gaits, joint angles are generated online through model predictive control (MPC), with multiple cost terms and constraints designed to balance task performance and motion safety. A small amplitude sinusoidal excitation is incorporated into the gait generation process to promote lateral undulatory characteristics while maximizing gait autonomy, thereby enhancing locomotion efficiency and stability. Furthermore, to address direction deviations induced by joint locking, an approach based on principal component analysis is adopted to estimate the principal axis of the robot for direction correction. Experiments with various joint locking angles validate the effectiveness of the proposed method in achieving path following control under these failure conditions. Interestingly, the automatically generated gaits are consistent with intuitive manual adjustment in similar situations. Note to Practitioners—Joint locking is a common failure mode in articulated robotic systems under sustained loading, and in snake robots it can significantly distort body configuration and impair path following performance. Experimental observations indicate that the impact strongly depends on the locking location: terminal joints mainly induce heading deviation, whereas middle joints lead to more severe instability and tracking degradation. Such effects are difficult to handle with predefined gait patterns due to their limited adaptability. The proposed framework addresses this challenge by generating joint motions online, allowing the system to reconfigure its shape using the remaining functional joints. In practice, appropriate motion constraints are important to ensure safe and physically consistent behaviors, while a small excitation helps preserve effective undulatory locomotion. A lightweight geometry-based direction correction further improves motion reliability under asymmetric configurations.