Robot confidence modeling and role change in physical human-robot collaboration
- Publication Type:
- Thesis
- Issue Date:
- 2019
Open Access
Copyright Clearance Process
- Recently Added
- In Progress
- Open Access
This item is open access.
In physical Human-Robot Collaboration, the human is generally in control of the interaction while the robot provides assistance to its human co-worker. However, with the increasing level of intelligence of robot co-workers, peer-to-peer interaction is expected and believed to be an ideal approach to collaboration between the human and the robot in a collaborative activity. In the peer-to-peer collaboration, the human and its robot co-worker would observe each other’s actions and intervene if one detects changes of their counterpart during the interaction which could negatively impact the task. Current research on safe pHRC only considers role change to be initiated from the human’s perspective, not from the robot’s perspective. This thesis aims to address three research challenges in pHRC: the robot’s perception of its human co-worker during pHRC, modeling the robot’s confidence in its human co-worker and how a robot would decide whether and when it should intervene (by taking control) in its human co-worker’s actions during pHRC.
This research first developed effective methods that enable the robot’s perception of its human co-worker during pHRC. The human’s grasping pattern and grasping strength on a handlebar, the commonly used interface in pHRC, are used by the robot to identify the orientation of the human co-worker’s hand and monitor the human’s reaction to unexpected events. A method for identifying the human’s hand orientation and detecting the human’s reaction to unexpected events was developed by analyzing the human’s grasping pattern and grasping strength.
The thesis then explored how the robot’s confidence in its human co-worker during pHRC can be modeled. A novel robot confidence framework was developed for modeling the robot’s confidence using the robot’s perception of the human’s performance. The framework was evaluated in a number of pHRC case studies where a robot and its human co-worker worked collaboratively.
Finally, this thesis explored how the robot’s confidence in its human co-worker can be used to decide whether and when the robot should initiate a role change. A confidence-based role change method was developed. Experimental verification of the role change method was conducted in a collaborative grit-blasting operation between a human and a robot. The results demonstrated that the method successfully identified the points during a pHRC where the robot should initiate a role change and take the control away from its human co-worker.
Please use this identifier to cite or link to this item: