Full text loading...
-
Research on Quadruped Crawling Robot Control with Finite-Time Observer in Continuous Convex Terrain
-
-
- 26 Feb 2024
- 11 Jun 2024
- 02 Oct 2024
Abstract
In recent years, observers have been crucial for controlling quadruped crawling robots, especially for feed-forward compensation in complex terrains. However, their tracking performance in continuous convex terrains requires optimization.
In order to improve the mobility of quadruped crawling robots by reducing posture adjustment time in continuous convex terrain, a finite-time observer with integration elements and optimized gait planning is proposed.
First, a buffer phase is introduced into the tripod gait planning to adapt to terrain changes, and the velocity error of the robot's center of mass is analyzed. Second, a coupled robot dynamics model is developed and the disturbance component structure is derived to ensure accurate estimation of the system state, especially in maintaining stability in the face of external disturbances. Finally, the auxiliary variable of the observer is based on the velocity error, with disturbance estimation terms combining power and sign functions, and the cumulative error is introduced by integrating the adjustment functions to improve the tracking performance.
The proposed observer reduces the maximum estimation error at knee and hip joints by 0.63 and 0.425 degrees, respectively, compared to a non-integrated observer. The integral of the absolute error during leg swing is 17.44% to 35.04% of the latter's, and the integral of the squared deviation error is 1.90% to 8.38%. These results demonstrate that the proposed observer can track the state information of the robot with greater accuracy.
The proposed finite-time observer significantly improves motion control for robots in continuous convex terrain and offers insights for enhancing stability in complex environments, which is expected to address the challenges of other robotic manipulator systems in overcoming the effects of external disturbances.