| Description |
Admittance-type robotic devices are commonly used to complete tasks that require a high degree of precision and accuracy because they appear nonbackdrivable to many disturbances from the environment. Admittance-type robots are controlled using admittance control; a human interacts directly with a force sensor mounted to the robot, and the robot is computer-controlled to move in response to the applied force. The experiment herein was conducted to determine under which operating conditions human velocity control is optimized for admittance devices that are controlled under proportional-velocity control, and to determine the degradation in control under nonoptimal conditions. In this study, the desired velocity of the device was shown on a visual display. The desired velocity was shown with a scaling factor from the actual velocity of the device because the device often moved at velocities too slow to perceive visually. The admittance gain, ka, desired velocity, Vd, and the visualization scale factor, S were tuned to adjust the user's experience when interacting with an admittance device. We found that in velocity-tracking tasks, scaling the visual feedback only has a significant effect on performance for very slow desired velocities (0.1mm/s), for the range of velocities tested here. In this thesis, we give evidence that there exists a range of velocities and forces within which humans optimally interact with admittance-type devices. We found that the optimal range of velocities is between 0.4mm/s and 1.0mm/s, inclusive, and the optimal range of forces is between 0.4 N and 4.0 N, inclusive. To ensure optimal velocity-control performance, the admittance gain should be selected such that the desired velocity and target force remain within their respective optimal ranges simultaneously. We also found that on average subjects moved faster than the desired velocity when the desired velocity was 0.1 mm/s and subjects were slower than the desired velocity when it was higher than 0.4 mm/s. For each admittance gain there is a different threshold velocity at which velocity-control accuracy is optimal in the aggregate. If the device operates at a velocity that is faster or slower than the threshold velocity the operator will tend to lag or lead the desired velocity, respectively. |
