Vertical Jumping for Legged Robot Based on Quadratic Programming

Area of Science:

• Robotics
• Control Systems
• Biomechanics

Background:

• Legged robots face challenges in controlling dynamic jumping motions.
• Existing control schemes struggle with over-constrained objectives in the stance phase and underactuation in the flight phase.

Purpose of the Study:

• To introduce a novel control framework for cyclic vertical jumping in planar legged robots.
• To address limitations in controlling both stance and flight phases of legged locomotion.

Main Methods:

• Formulation of the jumping problem within a quadratic-programming (QP) framework.
• Implementation of a hierarchical quadratic programming (HQP) control strategy for strict task prioritization.
• Incorporation of split dynamic equations to manage robot posture during the flight phase.

Main Results:

• The HQP controller successfully prioritized the center of mass (CoM) during the stance phase.
• The flight phase controller was validated in simulations, restricting robot posture.
• Maximum CoM errors in x and y directions were 0.47 cm and 0.82 cm, respectively.

Conclusions:

• The proposed QP-based framework with HQP control enables robust cyclic vertical jumping for legged robots.
• The integrated approach effectively manages coupled dynamics in the stance phase and underactuation in the flight phase.
• This method offers a promising solution for advanced legged robot locomotion control.