Enabling Autonomous Grasping with the Jaco Arm

This case study outlines the implementation of autonomous grasping for assisting in activities of daily living using the Kinova Jaco arm. The project aimed to develop a method for performing general grasping tasks autonomously, addressing limitations of teleoperation. The approach involved adapting the MoveIt Deep Grasps demo, which utilizes neural networks for grasping simulations, to work with the Jaco arm. Key objectives included generating grasp poses, enabling hardware functionality, and real-time object identification. The project successfully integrated live object detection using a RealSense camera, which provided RGB and depth images for grasping tasks. The final pipeline allowed users to specify objects for the Jaco arm to grasp, utilizing YOLO for object detection and a convolutional neural network for grasp pose generation. The project demonstrated the potential of the Jaco arm in assistive technology, highlighting its effectiveness in improving the quality of life for users through autonomous robotics.