Learning from Demonstration framework for Co-robots

Distributed Autonomous System Laboratory

Experimental set up for 1/14th scaled fully functional Excavator (a co-robot), which guides a novice operator to perform truck loading

Experimental setup for 1/14th scaled fully functional Excavator (a co-robot) guiding a novice operator to perform truck loading.

Collaborators and Students: 

Harshal Maske, Milecia Matthews, Allan Axelrod

Dr. Crick, Dr. Pagilla

Project Description: 

Human presence is indispensable for many co-robotic applications (robots that work with Human) such as construction robots, surgical robots, which is due to the safety critical nature of these real world tasks combined with their dynamically changing environment. Design of Intelligent Co-robots is likely to take the center-stage in coming decades. These co-robots are envisioned to perform complex tasks efficiently and optimally in collaboration with humans, as well as provide guidance to novice operators to accomplish complex tasks.

Video: Skill-transfer for Excavators Figure:  Excavator dumping over the truck

 

Derivative papers: 
  1. Maske, H.; Matthews, M.; Axelrod, A.; Mohamadipanah, H.; Chowdhary, G.; Crick, C. & Pagilla, P. Collaborative goal and policy learning from human operators of construction co-robots, Neural Information Processing Systems, 2014.