Skip to main content

News

Vulcan 1

The history of the Vulcan 1, as told by Dr. Ben Kuipers

Dr. Kuipers has worked on the representation and learning of commonsense knowledge of space for many years, including building robots to explore environments and create their own cognitive maps.  This capability seems foundational for many kinds of intelligent behavior, and the Vulcan 1 served as a project for exploring and demonstrating genuine value.

David Miller (now Professor Emeritus at the University of Oklahoma), had created the KISS Institute for Practical Robotics. Dr. Kuipers was impressed by one of their projects, a robot wheelchair, designed and built by Dave.  

Dr. Kuipers purchased the original version of Vulcan 1 robot wheelchair from the KISS Institute with funds from this grant in 1997:

  • National Science Foundation, CISE Research Instrumentation Program. “CISE Research Instrumentation: Robotics Equipment for Research on Assistive Intelligence.” (CDA-9617327.) 2-1-1997 to 1-31-2001.

At the time (probably 1995-96), Dr. Kuipers was teaching a course in which students built small robots out of Legos and programmed them with a micro-controller (the 6.270 board from MIT).  Through a friend who worked a UT office supporting people with disabilities, Dr. Kuipers invited a UT student who is paraplegic and uses a powered wheelchair full time, to give a guest lecture in his class.  Among other things, the student told the class that an intelligent robotic wheelchair could be of great practical value to himself, if it could keep him from bumping into the walls and furniture in his apartment, and losing his damage deposit. It seemed to Dr. Kuipers that this would be an easy starting point that could be extended to much more intelligent behavior if the robot could simultaneously learn a cognitive map of its environment from its own experience.

It was also clear, from that original guest lecture, that the autonomy must belong to the human wheelchair driver, with the robot in a subordinate role, doing the bidding of the human.  This contradicts the common idea of an "autonomous robot".  Any autonomy the robot has is temporarily delegated from the human, and that delegation can be changed or revoked at any time, for any reason.  This also raises a number of important research issues in human-robot interaction, which Dr. Kuipers described in several proposals and a few papers, and which still deserve a lot more serious attention.

Over the years after that,  many of the components were replaced, improving the wheelchair's sensing and control. This led to a vision of the basic powered wheelchair (with a carefully engineered, robust, and refined design by a company serving people with disabilities), controlled by a "robot nervous system" that would be added.  The sensors are attached to the wheelchair chassis, and to a computer placed somewhere on the wheelchair, sending motor commands to the wheelchair's motors.  In Vulcan 1, the computer was a "tower" style computer "backpack" behind the seat of the wheelchair.

Vulcan became the "workhorse" robot for robotics research at UT Austin.  The researchers tried hard trying to raise funds to start providing mobility services to actual people with disabilities, and got quite a bit of enthusiasm from NIH, but never quite enough to get funded.  Nonetheless, Vulcan 1 was a very useful platform for cognitive mapping research, and its potential applications were inspiring.

Dr. Kuipers has been the principal investigator for this work.  Other faculty collaborators include Brian Stankiewicz (UT Psychology) and Raul Longoria (UT Mechanical Engineering), and he has received a lot of inspiration from discussions with UT CS colleagues including Ray Mooney, Peter Stone, Bruce Porter, and Dana Ballard.

A number of UT students worked on projects with Vulcan, often publishing papers and dissertations describing their work. These students include:  Bill Gribble, Rob Browning, Mike Hewett, Emilio Remolina, Aniket Murarka, Matt MacMahon, Joseph Modayil, Patrick Beeson, Shilpa Gulati, Mohan Sridharan, and Changhai Xu.  

You can find a list of papers and projects (from both UT and UM) along with several videos (made with Vulcan 2 at UM) on the following website:

     https://web.eecs.umich.edu/~kuipers/research/wheelchair/

Dr. Kuipers moved from the University of Texas at Austin to the University of Michigan in Ann Arbor at the end of 2008, when his wife became Dean of the UM School of Social Work.  He wanted to continue his Intelligent Wheelchair research, so his start-up package at UM CSE included funds for a new powered wheelchair base.  His team designed and implemented a new "robot nervous system" for perception, cognition, and control.  Vulcan 2 includes a number of engineering improvements, and is capable of higher speeds and better maneuverability, but it is faithful to the same vision that inspired Vulcan 1.

As of 2023, Dr. Kuipers is now about a year and a half from retirement from the University of Michigan, and he is ramping down his Intelligent Wheelchair project.  However, two other researchers at the University of Michigan are just beginning a new robotic wheelchair project, and Dr. Kuipers have begun a collaboration with them to transfer as much of the project's technology as will be helpful to them.

 

Additional links:

About Dr. Kuipers

Vulcan 2 Video