Advisor

• Jane Li (zli11@wpi.edu)Assistant Professor with Robotics Engineering and Mechanical Engineering
• Jie Fu (jfu2@wpi.edu)Assistant Professor with Robotics Engineering and Electrical & Computer Engineering
• Craig Putnam (cbputnam@wpi.edu)Associate Director and Instructor with Robotics Engineering and Computer Science

Academic Year: 2017-2018

Major: RBE, ME, CS, ECE

Project Members:

• Emrick, Alexandra K. (akemrick@wpi.edu)
• Gudenrath, Sophia A. (sagudenrath@wpi.edu)
• Eldredge, Elijah J. (ejeldredge@wpi.edu)
• Cupo, Dominic J. (djcupo@wpi.edu)
• Hatfalvi, Mary T. (mthatfalvi@wpi.edu)

Project Goal

This project serves as the preliminary work towards a shared autonomous system for motion learning and control, so that a manipulator robot can intelligently, reliably and efficiently assist novices to perform the shoe-sewing tasks to the level of an expert worker. In shared autonomy, novice worker will teleoperate the assistive manufacturing robot, instead of commanding the robot to autonomously perform the tasks based simple sets of input. The assistive robot will augment the motor skills of the novice user, including from compensating tremors and collision detection and avoidance, to suggesting and guiding optimal operation, automating repetitive procedures based on edited and refined demonstration. As a one-year MQP project, this project will implement several fundamental autonomous functions of this shared autonomous system, including:

• Identifying the critical sewing features (e.g. landmarks and edges for fitting the parts) using vision sensing
• Path and trajectory planning and optimal control for stitching in 2D, based on motor skills learned from expert workers. 3D cameras maybe considered depending on the project budget. A typical commercial 3D camera (e.g. Kinect v2) can render point cloud of SD quality at frequency of 30 Hz with cm-level accuracy.
• Develop grippers appropriate for shoe-sewing tasks (optional). Investigation and implementation of gripper design, as well as vision-based motion planning and control, can be an independent MQP project.

The team is expected to work closely with other master students through different phases of the project.

Project Description

• Assistive robot for shoe-sewing tasks

Many manipulation tasks in manufacture, for example shoe-sewing, require very fine motor skills that can only be acquired after extensive training and practice. The task performance depends heavily on the expertise of workers, rather than the efficiency and accuracy of the sewing machines. Therefore, we propose to develop intelligent manufacturing assistive robot control system can (1) acquire the expertise of experience workers by learning from demonstration, and (2) convey such knowledge to the algorithms of motion planning and control, which only requires minimal and necessary guidance and correction from workers. The proposed intelligent control system takes advantages of the synergistic cooperation of human and robot, to lower the training curve while improve ergonomics and performance. Furthermore, this system also intends to relieve manufacture workers from tedious and repetitive tasks that demands intensive mental efforts. Further research such how to handle deformable objects and how to coordinate manipulation and perception are proposed as graduate-level projects.

Required Skills

• Robotic kinematics and dynamics, programming skills (Matlab, C/C++, Python)
• Experience with manipulator robots, vision sensing and machine learning algorithms can be a plus