Robots Performing Day-to-day Tasks

In our STAIR project we want to build a robot capable of performing a variety of day-to-day tasks in home and office environments.  In the long term we want it to fetch and deliver items, show guests around the lab and clean up after a party.  For starters, we teach it to navigate hallways and open doors.



(Note: these videos have been sped up for better viewing)


Fully autonomous experiments including navigation and manipulation.  In these experiments the robot relies on a single laser sensor.

· Visiting a conference room

· Stopping by printer room

· Entering an office


From our earlier work:

Opening doors using contact sensing (2005-11-01).  In this video the robot locates the door handle by touching it.  It does not use any other sensors.  Only door handle probing and manipulation are autonomous.  Motion of the mobile platform is controlled with a joystick.



Relevant publications:


Probabilistic Mobile Manipulation in Dynamic Environments, with Application to Opening Doors.  Anna Petrovskaya and Andrew Y. Ng.  International Joint Conference on Artificial Intelligence (IJCAI), Hyderabad, India, 2007. [pdf]

Bayesian Estimation for Autonomous Object Manipulation Based on Tactile Sensors.  Anna Petrovskaya, Oussama Khatib, Sebastian Thrun, and Andrew Y. Ng. IEEE International Conference on Robotics and Automation (ICRA), Orlando, Florida, 2006. [pdf]  Also poster version from WiML06 [ppt]