Joint Position Control of aWalking Humanoid Robot with MPC Controller
No Thumbnail Available
Date
2024-06
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
With much of the industrial processes in Ethiopia relying on human labor, people are
obliged to perform hazardous and monotonous tasks, such as lifting heavy objects, working
in contaminated environments, and engaging in repetitive activities. Unfortunately,
this places human lives at risk and exposes their health to potential harm. Moreover,
these conditions also negatively impact the industry itself, leading to decreased production
quantity and quality. The main objective of this research is to model and simulate
a humanoid robot with each leg having 4 degrees of freedom (DOF) and each arm having
4 DOF, enabling it to carry heavy loads and move to different locations. To control
the joint position of the robot, an MPC has been implemented and a comparison
with a LQR has been done to evaluate the performance. Particle swarm optimization
technique has been utilized to tune parameters of the controller yielding better performance.
The approach first started with a thorough understanding of problem followed
by a 3D modeling of a humanoid robot model in Solidwork. Further with the exported
model in Simulink, different sets of angles were assigned to the robot joints so as to
imitate human motion. To provide transition from a departure to destination, a sets of
way points have been defined in MATLAB and the controller’s ability to transit the dynamic
system in a stable manner have been tested. The pure pursuit controller has been
implemented to track the path, allowing the humanoid robot to navigate through the defined
set of points. The implemented MPC has been found to provide efficient trajectory
tracking as compared to LQR. The comparison have been further analyzed by calculating
the peak error value, where the MPC Controller provided better tracking performance
with peak error values of 0.13 in shoulder joint and 0.18 in both hip and knee
joint as compared to error values of 0.38 and 0.58 obtained while implementing LQR.
Description
Keywords
Humanoid Robot, MPC, LQR, Pure pursuit, PSO