An error occurred while loading the PDF.
The Elliott and Connolly Benchmark: A Test for Evaluating the In-Hand Dexterity of Robot Hands
Voiceover
Presenter(s)
Chao Li
Programs/Groups
Sigma Xi Poster Competition
Abstract or Description
Achieving dexterous manipulation with robot hands is an extremely challenging problem, in part due to current limitations in hardware design. We believe that one of the bottlenecks hampering the development of improved hardware for dexterous manipulation is the lack of a standardized benchmark for evaluating the dexterity of robot hands. In order to address this issue, we establish a new benchmark for evaluating dexterity: the Elliott and Connolly Benchmark. This benchmark consists of 13 manipulation patterns which cover the most important aspects of in-hand dexterity. We define qualitative and quantitative metrics for evaluation of the benchmark, and provide a detailed testing protocol. Our benchmark test on the CMU Foam Hand II provides a quantitative way to access the dexterity of robotic hands. Our results also shown that the human hand might not be an optimal design for all types of dexterous manipulations.
Mentor
Nancy Pollard
Comments
Removed4 years ago
This comment has been deleted.
Removed4 years ago
This comment has been deleted.
Chao Li4 years ago
We also provide additional media to demonstrate how each manipulation is performed. Robot Hand: (https://www.youtube.com/watch?v=PXa9MyvtvmQ); Human Hand: (https://www.youtube.com/watch?v=HBEgVw83_pY)
•
Angela Li4 years ago
Does the camera capture a pixel distance that is directly proportional to the actual distance moved in the real world?
• • 1 comment
Chao Li4 years ago
That is a great question! The transformation between camera space to world space is not linear due to lens distortion arising from the curvature of the camera lenses. We performed camera calibration using a checkerboard to get the intrinsic camera parameters. These parameters can then be used to correct for lens distortion and flatten the images. Now, we have a linear transformation between the flattened image and world space. So yes, we did account for the non-linearity that arises from the curvature of lenses that are inherent in cameras!