From industrial robots to nursing robots, object manipulation has become a growing area of robotics research. This Major Qualifying Project explores methods of teleoperation through the use of a wireless data glove able to detect multiple degrees of freedom. Our project also explored methods for autonomous control. We developed a computer vision model by integrating two state-of-the-art Mask Region Convolutional Neural Networks (Mask-RCNN) models to create a final model for determining both object location and grasp angle. This modeling allows the Baxter Robot to autonomously detect and reach towards the object. Using learning by demonstration, the robot can learn how to grasp and manipulate said objects.

• This report represents the work of one or more WPI undergraduate students submitted to the faculty as evidence of completion of a degree requirement. WPI routinely publishes these reports on its website without editorial or peer review.

Creator

• Bunnag, Bhon
• Ditthapron, Apiwat
• Buckley, Colin Matthew
• Miles, Rebecca Eileen

Publisher

• Worcester Polytechnic Institute

Identifier

• E-project-042019-150243

Advisor

• Li, Zhi
• Bitar, Stephen John
• Fichera, Loris

Year

• 2019

Date created

• 2019-04-20

Resource type

• Major Qualifying Project

Major

• Computer Science
• Electrical & Computer Engineering
• Robotics Engineering

Rights statement

• In Copyright