A Whole-Body Coordinated Motion Control Method for Highly Redundant Degrees of Freedom Mobile Humanoid Robots

Abstract

Humanoid robots are becoming a global research focus. Due to the limitations of bipedal walking technology, mobile humanoid robots equipped with a wheeled chassis and dual arms have emerged as the most suitable configuration for performing complex tasks in factory or home environments. To address the high redundancy issue arising from the wheeled chassis and dual-arm design of mobile humanoid robots, this study proposes a whole-body coordinated motion control algorithm based on arm potential energy optimization. By constructing a gravity potential energy model for the arms and a virtual torsional spring elastic potential energy model with the shoulder-wrist line as the rotation axis, we establish an optimization index function for the arms. A neural network with variable stiffness is introduced to fit the virtual torsional spring, representing the stiffness variation trend of the human arm. Additionally, a posture mapping method is employed to map the human arm potential energy model to the robot, enabling realistic humanoid movements. Combining task-space and joint-space planning algorithms, we designed experiments for single-arm manipulation, independent object retrieval, and dual-arm carrying in a simulation of a 23-degree-of-freedom mobile humanoid robot. The results validate the effectiveness of this approach, demonstrating smooth motion, the ability to maintain a low potential energy state, and conformity to the operational characteristics of the human arm.

Keywords: dynamic movement primitives; humanoid motion; mobile humanoid robot; whole-body motion control.

Figure 1

Whole -body coordinated motion control architecture for mobile humanoid robot.

Figure 2

Illustration of grasping experiments at different heights.

Figure 3

Violin plot of elbow and wrist height variation during arm movement. (a) Changes in wrist and elbow height in the desktop grasping experiment. (a) Changes in wrist and elbow height in the lift-arm experiment.

Figure 4

Illustration of arm angle.

Figure 5

Human-robot posture mapping process.

Figure 6

Neural network for virtual stiffness fitting. (a) Neural network architecture used for virtual torsion spring fitting process. (b) Loss variation during training process.

Figure 7

Mobile humanoid robot model.

Figure 8

Experimental simulation scenario for the mobile humanoid robot.

Figure 9

Trajectory generalization process of the robot’s right arm. (a) Generalized trajectory. (b) Execution process of the robot’s generalized trajectory. Where stages (a) to (d), (e) to (h), and (i) to (l) respectively indicate that the robot performs three different generalization trajectories.

Figure 10

Obstacle avoidance process in operational space trajectory. (a) Obstacle avoidance trajectory. (b) Robot obstacle avoidance trajectory execution process.

Figure 11

Independent dual-arm grasping and placement process. Stages (a–l) represent the turning and placing process after the robot grasps the object.

Figure 12

Joint velocity of the mobile humanoid robot. (a) Left arm joint velocity. (b) Right arm joint velocity. (c) Waist joint velocity. (d) Mecanum wheel velocity.

Figure 13

Energy variation diagram. (a) Total potential energy variation curve of the arm. (b) Trend of virtual torsional spring elastic potential energy of the arm. The blue dots represent the actual data, and the red dotted line represent the corresponding trend.

Figure 14

Comparison of our proposed method with the pseudo-inverse method. (a) Cases of motion planning failure using the pseudoinverse method. (b) Motion Effect of Pseudoinverse Method. (c) Motion Effect of the Proposed Method.

Figure 15

Illustration of primary-secondary arm task-space trajectories.

Figure 16

Dual-arm coordinated grasping motion process.

Figure 17

Joint velocity of the mobile humanoid robot. (a) Left arm joint velocity. (b) Right arm joint velocity. (c) Waist joint velocity. (d) Mecanum wheel velocity.