Purdue University Mark

Purdue University

Richard Voyles' Self-Adaptive Systems

Reconfigurable and Autonomic Computing

As Site Director of the NSF Center for Robots and Sensors for the Human Well-Being (RoSe-HUB), Prof. Voyles RoSe-HUB is co-sponsoring some of this work. This site also reflects work as a participant in the National Science Foundation grant 0421159.

We are developing a reconfigurable computing platform for Heterogeneous Wireless Control Networks to support the TerminatorBot, the Dexterous Hexrotor, MOTHERSHIP and other novel robotic mechanisms for search-and-rescue. This wireless, distributable node is being used as the backbone of high-performance wireless sensor networks research, wireless video sensor networks research, heterogeneous sensor netowrks, and static and dynamic reconfigurable computing research.

Autonomic Computing is a term derived from the biological description of our Autonomic Nervous System. The autonomic nervous system is that part of the nervous system that maintains homeostasis, or the internal status of the system (human body, in this case). Temperature, hydration level, blood pressure, etc. are elements of the internal status of the human body. These are self-regulating in the face of varying environmental conditions.

Likewise, we are interested in self-regulation of robots through self-reconfiguration of hardware and software -- an autonomic system. One example of our autonomic robots is a robot that can identify terrain based on gait effects, and then modify the gait to maintain some level of locomotion efficiency. (See Larson, Demir, and Voyles, 2005, below.)


Real-time software architecture for self-adaptive systems.

ReFrESH - Reconfigurable Framework for Embedded systems both Software and Hardware - is a multi-layered software middleware architecture over a real-time kernel that provides design-time and run-time infrastructure for the construction of self-adaptive systems. Like real-time middleware made working with real-time operating systems easier, middleware for fault tolerance and self-adaptivity is making systems that can adapt to and learn from changes in their environment a mangeable task.


The TerminatorBot is a search-and-rescue robot that is unique in its ability to manipulate objects and crawl over very difficult terrain in such a small package. Inspired by the final scene of the original Terminator movie, this sode can-size robot is able to manipulate objects with its limbs and locomote by dragging itself with the same limbs. Below are pictures of the first two 75mm-diameter prototypes of the TerminatorBot. A third has actuated claws for climbing down ropes. (Climbing action has yet to be perfected. Stay tuned for videos future...)

The TerminatorBot is intended to be a manipulative node in a heterogeneous fabric for ubiquitous computing based on the reconfigurable computing node described here. Nodes in this computational fabric will contain many of the capabilities needed for a robot: computation, wireless communication, sensing, and manipulation.

Pair of TerminatorBots
Two TerminatorBots collaboratively navigating on rock and wood chips.

As part of a workshop sponsored by the NSF R4 program and the NSF Safety, Security, and Rescue Research Center, the TerminatorBot is shown here at the rubble pile at the Lakehurst Naval Air Station with New Jersey Task Force 1.

TerminatorBot at Lakehurst
TerminatorBot after a drop into a vertical sewer pipe.


Robotics Sources

Distributed Robotics Source List

Semiconductor Manufacturers

Copyright: © 2007 by Richard M. Voyles

rvoyles [at] purdue [dot] edu

Purdue University, West Lafayette, IN 47907, (765) 494-4600