A Model-Free Solution for Stable Balancing and Locomotion of Floating-base Legged Systems

Author(s):  
Emmanouil Spyrakos-Papastavridis ◽  
Jian S. Dai
2021 ◽  
Vol 13 (2) ◽  
Author(s):  
Emmanouil Spyrakos-Papastavridis ◽  
Jian S. Dai

Abstract This paper attempts to address the quandary of flexible-joint humanoid balancing performance augmentation, via the introduction of the Full-State Feedback Variable Impedance Control (FSFVIC), and Model-Free Compliant Floating-base VIC (MCFVIC) schemes. In comparison to rigid-joint humanoid robots, efficient balancing control of compliant bipeds, powered by Series Elastic Actuators (or harmonic drives), requires the design of more sophisticated controllers encapsulating both the motor and underactuated link dynamics. It has been demonstrated that Variable Impedance Control (VIC) can improve robotic interaction performance, albeit by introducing energy-injecting elements that may jeopardize closed-loop stability. To this end, the novel FSFVIC and MCFVIC schemes are proposed, which amalgamate both collocated and non-collocated feedback gains, with power-shaping signals that are capable of preserving the system's stability/passivity during VIC. The FSFVIC and MCFVIC stably modulate the system's collocated state gains to augment balancing performance, in addition to the non-collocated state gains that dictate the position control accuracy. Utilization of arbitrarily low-impedance gains is permitted by both the FSFVIC and MCFVIC schemes propounded herein. An array of experiments involving the COmpliant huMANoid reveals that significant balancing performance amelioration is achievable through online modulation of the full-state feedback gains (VIC), as compared to utilization of invariant impedance control.


Author(s):  
Chunhong Zheng ◽  
Yuxin Su ◽  
Paolo Mercorelli

Abstract This paper revisits the problem of global asymptotic positioning of uncertain motion systems subject to actuator constraint and friction. A simple model-free saturated control is proposed by incorporating a relay term driven by position error into proportional-derivative (PD) methodology. Lyapunov's direct method is employed to prove global asymptotic positioning stability. The appealing advantages of the proposed approach are that it is conceived within the framework of saturated PD (SPD) control methodology with intuitive structure and absence of modeling parameter and embeds the whole control action within a single saturation function. Benefitted from these advantages, the proposed approach omits the complicated discrimination of the terms that shall be bounded in several saturation functions of the commonly used design and permits easy implementation with an improved performance. An additive feature is that the proposed control has the ability to ensure that the actuator constraint is not breached and assures global asymptotic positioning stability in the presence of unknown friction. Numerical simulations and experimental validations demonstrate the effectiveness and improved performance of the proposed approach. The proposed approach provides a model-free solution for fast transient and high-precision steady-state positioning of uncertain motion systems subject to unknown friction and actuator constraint.


IEEE Access ◽  
2020 ◽  
Vol 8 ◽  
pp. 112253-112265
Author(s):  
Jingjing Cui ◽  
Zhiguo Ding ◽  
Yansha Deng ◽  
Arumugam Nallanathan ◽  
Lajos Hanzo

2020 ◽  
Vol 43 ◽  
Author(s):  
Peter Dayan

Abstract Bayesian decision theory provides a simple formal elucidation of some of the ways that representation and representational abstraction are involved with, and exploit, both prediction and its rather distant cousin, predictive coding. Both model-free and model-based methods are involved.


Sign in / Sign up

Export Citation Format

Share Document