Assistive Robotics

Principal Investigator: Gary Lin

Co-PIs: Jeremy Fishel, Gerald Loeb

Personal assistive robots are essentially lightweight versions of industrial telerobots, which have been used for decades to enable human workers to handle dangerous materials (e.g. nuclear fuel) or to work in dangerous environments (e.g. under seas, in outer space, on battlefields, etc.). Such work now requires a great deal of training and is tedious to perform and prone to errors because operators lack normal tactile sensing and feedback, relying entirely on slow and limited visual perception.

Project Objective

Intuitive and simple control interfaces are particularly important for the users of personal assistive robots. Our work seeks to develop and validate haptic controllers that improve the performance of these robots. We have previously developed simplified controllers to control the fingers of the Barrett to stop each finger on contact allowing an object to be grasped without being knocked over. Future research will integrate biomimetic control strategies into algorithms that users will access via a typical joystick controller for an assistive robot. We will validate the system design with human operators using our industrial robotic platform, which can emulate the performance of the various personal assistive robots now available or coming onto the market. The resulting control system will then be integrated onto the specific robotic platforms of our industrial partners

Specific Aims

  1. Implement human-like tactile sensing and low-level grip adjustment behavior with high-level commands from the human operator
  2. Develop continuous on-line grip-adjustment algorithms that handle transitions during lifting and placing objects that have various friction coefficients
  3. Develop voluntary grasping solutions on assistive robotic platforms
  4. Assess task performance with and without haptic controllers

Media:

Conceptual picture of the future robotic system grasping a container out of the fridge As the fingers close around an object, each finger is stopped when contact is made, preventing the object from being knocked over as the fingers close around it. Conceptual schematic demonstrating how contact detection can be used to detect and reorient robotic fingers for stable grasp.

Publications


Analytical point of contact, force and torque calculations for the BioTac Lin, C.H., Fishel, J.A., Loeb, G.E. Unpublished. 2013 Upcoming
Grip control using biomimetic tactile sensing systems Wettels, N., Parnandi, A.R., Moon, J.-H., Loeb, G.E. and Sukhatme, G.S. IEEE/ASME Trans. Mechatronics, 14(6):718-723. 2009 Download PDF
Site Map || Job Opportunities || Ordering Information || Contact