A Disturbance Rejection Control Method Based on Deep Reinforcement Learning for a Biped Robot

Liu, Chuzhao; Gao, Junyao; Tian, Dingkui; Zhang, Xuefeng; Liu, Huaxin; Meng, Libo

doi:10.3390/app11041587

Cited by 6 publications

(3 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Most subsequent bipedal robot research studies [30][31][32] utilized the cart-table model and ZMP to establish a stable CoM trajectory (ZMP trajectory), and then applied this trajectory to the bipedal robot to design controllers and path-planning methods, while minimizing the interference of the external environment during locomotion. Even the latest deep-learning research [33] did not yield any breakthroughs in the dynamic balance of bipedal robots, and control processes still rely on planning the ZMP trajectory. This study combined the aforementioned methods to achieve realistic bipedal robot simulations of human walking gaits.…”

Section: Zero-moment Point (Zmp)mentioning

confidence: 99%

Control Design for an Exoskeleton System Based on the Human Gait

Lee,

Tsai,

Yen

et al. 2023

Preprint

View full text Add to dashboard Cite

Exoskeleton systems are wearable devices originally designed to enhance the physical abilities of humans, and subsequently for medical rehabilitation purposes. Impedance control is primarily used to achieve adaptability in exoskeleton control research. The concept of motion balance has not yet been applied to exoskeleton systems, which led us to propose a digital-twin concept, which operates in parallel with the operator. The digital twin is calculated by analyzing the dynamic characteristics of the entire human body, with the aim of determining the joint torques required to maintain balance and stability. If a discrepancy is detected between operator movements and the torques calculated using the digital twin, the system promptly provides assistance to ensure the safe operation of the exoskeleton. This study implemented the digital twin by utilizing the MATLAB Simscape modeling tool to construct a bipedal robot model that conforms to human body proportions and simulates real human postures. The zero-moment point was used as a reference point in the model for the interaction forces between the bipedal robot and the ground. We adopted the cart-table model to simplify the gait balance control of the bipedal robot model. The controller outputs an assistive force at the position of the center of mass to maintain balance during walking. Finally, we converted the assistive force into ankle joint outputs to enhance gait stability and ensure that the waveform and numerical outputs were closer to real human data found in the literature.

show abstract

Section: Zero-moment Point (Zmp)mentioning

confidence: 99%

Control Design for an Exoskeleton System Based on the Human Gait

Lee,

Tsai,

Yen

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…the optimal parametric strategy based on an inverted pendulum with flywheel used to plan the robot's motion, (b). the heuristic strategy to prevent the robot from bouncing and rolling over [27], and self-disturbance rejection control [28,29]. An example of CoM trajectory generation based on predictive control is shown in Fig.…”

Section: Com (Center Of Mass) Trajectorymentioning

confidence: 99%

Optimization of Humanoid Robot Leg Movement Using Open CM 9.04

Wajiansyah

Malani

Supriadi

et al. 2022

Journal of Robotics and Control

View full text Add to dashboard Cite

The Indonesian Robot Dance Contest (KRSTI) is a branch of the Indonesian Robot Contest (KRI) with the theme of dance. The robot used is a humanoid robot that can dance. Every year at the event, the provisions for robots constantly change, both the type of dance being demonstrated and the requirements for the robot's height. The taller the robot, the more difficult it is to control its walking movements because of the load it carries. This study uses a suitable algorithm to make the walking movement more natural and minimize the robot's falling. Human ROM data is used as a parameter for the range of motion of the servos that act as joints in the robot's legs. The algorithm created serves to determine the initial position of the angle on the servo to avoid the wrong initial movement position between one servo and another. The robot used is the Bioloid Robot’s leg Type A and uses OpenCM 9.04-A as the controller. The results showed that ROM on human feet could not be fully implemented on robot legs due to the robot's structure and the need for a robot that only relies on an algorithm to find the correct fulcrum to maintain balance. The comparison results show that the movement when walking on the ankle (ID servo 15) ranges from 749-567, while the ROM range is only between 580-512. When walking (servo ID 16), movement ranges from 460-291, while the ROM range ranges from 580-512.

show abstract

“…In recent years, researchers have been trying to use machine learning to develop biped robot balance controllers, as reported in Refs. [14,15].…”

Section: Introductionmentioning

confidence: 99%

Disturbance rejection for biped robots during walking and running using control moment gyroscopes

Zhou

Chen

et al. 2022

IET Cyber-Syst and Robotics

View full text Add to dashboard Cite

Keeping balance in movement is an important premise for biped robots to complete various tasks. Now, the balance control of biped robots mainly depends on the cooperation of various joints of the robot's body. When robots move faster, the adjustment allowance of joints is reduced, and the robot's anti-disturbance ability will inevitably decline. To solve this problem, the control moment gyroscope (CMG) is creatively used as an auxiliary stabilisation device for fully actuated biped robots and the CMG assistance strategy, which can be integrated into the biped's balance control framework, is proposed. This strategy includes model predictive control module, distribution module, and CMG precession controller. Under the command of it, CMGs can effectively assist the robot in resisting impact and returning to initial positions in time. The results of anti-impact simulation on the walking and running biped robot prove that, with the help of CMGs, the robot's ability to resist disturbance and remain stable is significantly improved. The cover image is based on the Original Article Disturbance rejection for biped robots during walking and running using control moment gyroscopes by Haochen Xu et al.,

show abstract

A Disturbance Rejection Control Method Based on Deep Reinforcement Learning for a Biped Robot

Cited by 6 publications

References 22 publications

Control Design for an Exoskeleton System Based on the Human Gait

Control Design for an Exoskeleton System Based on the Human Gait

Optimization of Humanoid Robot Leg Movement Using Open CM 9.04

Disturbance rejection for biped robots during walking and running using control moment gyroscopes

Contact Info

Product

Resources

About