Task-space separation principle: a force-field approach to motion planning for redundant manipulators

Area of Science:

• Robotics
• Neuroscience
• Biomechanics

Background:

• Human motor control exhibits complex postural synergies during redundant tasks.
• Existing models struggle to integrate postural control with task-space movement planning.
• Donders' law describes consistent head posture during eye movements, a key human motor synergy.

Purpose of the Study:

• To develop a computational framework for human-like motor planning in redundant manipulators.
• To implement and validate postural synergies, such as Donders' law, within a kinematic model.
• To solve the posture/movement problem by separating postural control from task-space motion control.

Main Methods:

• Framing the redundancy problem as a constrained optimization using Lagrange multipliers.
• Applying a novel task-space separation principle to Lagrange multipliers, separating posture and movement control.
• Utilizing nonlinear inverse optimization (NIO) to fit model parameters to human experimental data.

Main Results:

• Successfully applied the task-space separation principle to a redundant human wrist pointing task.
• Nonlinear inverse optimization captured and simulated human motor strategies with high fidelity.
• The framework effectively separated postural control (Donders' law) from task-space movement planning.

Conclusions:

• The proposed framework enables human-like motor planning in redundant robots by integrating postural synergies.
• The task-space separation principle offers a novel approach to solving the posture/movement problem in robotics.
• This work provides a method for creating more natural and efficient robotic movement controllers.