A Torque-actuated dissipative spring loaded inverted pendulum model with rolling contact and Its application to...

Area of Science:

• Robotics
• Mechanical Engineering
• Dynamical Systems

Background:

• The spring loaded inverted pendulum model with rolling contact (R-SLIP) is a foundational model for understanding legged locomotion.
• Advancements in robotic systems necessitate more sophisticated models that incorporate active control and dissipative elements.

Purpose of the Study:

• To introduce and analyze a novel torque-actuated dissipative spring loaded inverted pendulum model with rolling contact (TDR-SLIP).
• To compare the TDR-SLIP model with the R-SLIP model and investigate the impact of rolling contact.
• To evaluate the TDR-SLIP model's efficacy in controlling an empirical robot for stable locomotion.

Main Methods:

• Numerical analysis of stability using steps-to-fall and return map analyses.
• Variation of dimensionless parameters (torque, damping, spring constant, angles, speed) to assess dynamic performance.
• Development of a parameter equivalency method for comparing TDR-SLIP and R-SLIP models.
• Experimental validation using an empirical robot with different leg morphologies and control strategies (position-based vs. hybrid control).

Main Results:

• The TDR-SLIP model exhibits unique features due to its rolling contact, differentiating it from the R-SLIP model.
• Empirical robot experiments showed that compliant legs matched R-SLIP behavior, while mechanical legs matched TDR-SLIP behavior.
• Hybrid control strategy resulted in more stable robot locomotion, closely adhering to the model's dynamic profile.
• Stable robot motion aligned with the TDR-SLIP model dynamics significantly reduced leg energy cost during stance compared to flight.

Conclusions:

• The TDR-SLIP model provides a more advanced framework for analyzing and achieving stable legged locomotion.
• The integration of torque actuation and dissipative elements, particularly with rolling contact, enhances dynamic performance.
• Hybrid control strategies are effective in translating model dynamics to real-world robotic applications, improving stability and energy efficiency.