Object Detection with an Anthropomorphic Hand

Abstract

Dexterous robotic manipulation is a high-growth research field with substantial barriers to entry. Existing options for experimental hardware are prohibitively expensive and often lack the robust tactile sensing capabilities needed for research. Our project, sponsored by the MMINT Lab, aims to develop an accessible and environmentally perceptive anthropomorphic robotic hand for use as an advanced research platform. By integrating the “GelSlim 4.0”, a computer vision-based tactile sensor developed by the sponsor, we deliver dextrous manipulation capabilities beyond those of similar competitors. We release our 3D CAD files, modified electronics, ROS communication API and graphical control interface.