NSPG: An Efficient Posture Generator Based on Null-Space Alteration and Kinetostatics Constraints

Area of Science:

• Robotics
• Control Systems
• Computational Geometry

Background:

• Locomotion and contact planners for multi-limbed robots often reduce search space for performance.
• Posture generation is crucial for collision-free, stable robot configurations but is challenging in cluttered environments.

Purpose of the Study:

• To present a novel posture generator, the null-space posture generator (NSPG), for multi-limbed robots.
• To efficiently generate feasible, collision-free, and statically stable whole-body postures in complex scenarios.

Main Methods:

• The NSPG utilizes hierarchical inverse kinematics and contact force optimization.
• It exploits the null-space of the contact manifold to alter nominal postures while satisfying constraints.
• An adaptive random velocity vector generator explores the robot's workspace.

Main Results:

• The NSPG efficiently generates feasible configurations in short times, outperforming previous methods.
• Simulations with quadruped and biped robots demonstrated suitability for cluttered environments.
• The method achieved a twofold improvement in generating feasible configurations within the same time period.

Conclusions:

• The NSPG is a valid and generalizable method for robot posture generation.
• It offers significant performance improvements, especially in complex, cluttered scenarios.
• Real-world experiments on the CENTAURO platform confirmed its applicability.