Loading...
Advanced shared autonomy from controlled environments to the real-world for object telemanipulation
Bowman, Michael S.
Bowman, Michael S.
Citations
Altmetric:
Advisor
Editor
Date
Date Issued
2023
Date Submitted
Collections
Research Projects
Organizational Units
Journal Issue
Embargo Expires
Abstract
Successfully teleoperating a robot for an object manipulation task is often difficult and complex due to indirect visualization, indirect manipulation with the robot, and physical discrepancies between a human and robot hand. Shared control injects autonomy into an operator’s actions to overcome these issues. Even though the approaching/navigation process is one of the major components of teleoperation, the object manipulation task after approaching is essential and challenging but has not received enough research attention. Object telemanipulation requires more than simple position control, especially when considering real-world adoption. These real-world scenarios hold significant challenges: 1) human motion is ambiguous to decipher, 2) operators may solve the same task differently and there are at least 6 dimensions that they control, thus each dimension requires a different level of assistance to aid in the task's success, and 3) control policies are too strict to adapt under real-world scenarios. Standard shared control in navigation is insufficient to address these issues for telemanipulation, causing low success rates and feeling of a loss of control. Three objectives are developed to overcome these issues: 1) disambiguating the grasp motion goal with task level inference and determining the actions needed to satisfy it, 2) provide a dimension-specific shared autonomy to provide appropriate assistance for good task performance and maintain a feeling in control of the robot, and 3) provide an adaptive filter-based shared autonomy, so an operator’s corrective actions are not always rejected when differing from the autonomous agent’s policy. Overall, this dissertation presents novel shared control algorithms that can be used to improve a human operator’s control of a robotic manipulator to achieve a telemanipulation task that would otherwise be challenging or impossible for the user on their own.
Associated Publications
Rights
Copyright of the original work is retained by the author.