Online Stability in Human-Robot Cooperation with Admittance Control

Area of Science:

• Robotics
• Control Systems
• Human-Robot Interaction

Background:

• Designing compliant admittance controllers for human-robot interaction (HRI) requires ensuring stable and effective cooperation.
• Controller stability is challenged by stiff environments, often leading to conservative control gains that limit cooperation effectiveness.

Purpose of the Study:

• To propose a method for detecting unstable behavior in admittance controllers and stabilizing robots through online adaptation of control gains.
• To relax conservative gains and improve cooperation by considering the impact of variable admittance on operator effort.

Main Methods:

• Utilizing frequency domain analysis to develop an instability index based on high-frequency force signal oscillations.
• Implementing an adaptation scheme for admittance parameters to dynamically adjust control gains.
• Investigating two co-manipulation tasks: zero stiffness and stiff double-wall virtual environments.

Main Results:

• The proposed method effectively detects unstable behavior using frequency domain analysis.
• Online adaptation of admittance parameters successfully stabilizes the robot and relaxes conservative gains.
• Experimental validation with human subjects demonstrated improved cooperation in both tested environments.

Conclusions:

• The developed instability index and adaptation scheme enhance the stability and effectiveness of compliant admittance controllers in HRI.
• This approach allows for less conservative control gains, leading to improved human-robot cooperation, especially in challenging, stiff environments.