'

Plant cells for robot control

Why not apply lessons from nature to robotics? This is what did researchers from Japan and U.K. who built an interface between a plant and an omnidirectional hexapod robot. The interaction loop they realized between living plants and robots might lead to more biohybrid architectures.

Controlling robots in complex environments is not an easy task, but it would greatly increase their usages. Yet living organisms are facing complexity and successfully dealing with it. So why not apply lessons from nature to robotics? This is what did researchers from Japan and U.K. who built an interface between a plant and an omnidirectional hexapod robot. The interaction loop they realized between living plants and robots might lead to the integration of biological cells in other technological devices based on biohybrid architectures.

This research work has been recently presented at the "Biologically Inspired Approaches to Advanced Information" (Bio-ADIT 2006) conference held in Osaka, Japan, on January 26-27, 2006 under the name "Robot Control: From Silicon Circuitry to Cells."

This paper is published by Springer-Verlag GmbH in the proceedings of the conference, "Lecture Notes in Computer Science, Volume 3853." Here is a link to the abstract which was not enough to satisfy my curiosity.

Fortunately, one of the authors, Klaus-Peter Zauner, from the University of Southampton, made the full paper available online (PDF format, 13 pages, 1.15 MB).

This paper starts by looking at the biological paradigm and cellular information processing before turning to the characteristics of Physarum Plasmodia and to the Plasmodium properties used in the robot controller.

But the main part of the paper is about the robot control with Physarum circuits. Here is the introduction of this section.

In contrast to conventional information processing architectures, the infrastructure of a cell is in a continual state of flow. This dynamic replacement of components serves maintenance and facilitates structural reorganisation. Self-modification is thus inherent in cells. It would be very difficult to capture even part of this adaptive aspect in a purely artificial device based on current technology. By taking a hybrid approach that integrates a cell into a robot control architecture, however, we can experimentally investigate the interaction of a device that comprises autonomous, self-modifying, and self-maintaining components with its environment.

Here are some illustrations coming from this paper (Credits for images and captions: Soichiro Tsuda, Klaus-Peter Zauner, Yukio-Pegio Gunji).

First are some images of the Physarum circuit: "The plasmodium is patterned as six oscillators with star coupling by means of a negative plastic mask (A). Panel (B) to (D) show snapshots of the plasmodia growth. Shortly after preparation (B), 5 hours after preparation (C), and the fully developed circuit 10 hours after preparation (D)."

Plant cells to control robots: the Physarum circuit

Then here is "a tethered hexapod robot (A), used for its stance stability and gait flexibility. Each leg has only one degree of freedom and swings radial to the body (B). Whether a leg is in contact with the ground during its move will depend on the position of the other legs."

Plant cells to control robots: a tethered hexapod robot

Finally here is a diagram showing the cellular robot controller.

Plant cells to control robots: the cellular robot controller

In their conclusion, the authors mention that "the difference between natural and artificial information processors has received considerable attention in the theoretical literature," but that experimental approaches from an information processing perspective have been difficult to build.

Although our experiments are at an early stage, we expect that the biohybrid architecture presented above will open a path in this direction. From our experience so far, we conclude that the plasmodium of Physarum polycephalum is well suited to study device architectures based on autonomous components. At this stage many questions are open. We currently study the effect of the light input signals on circuits to gain a better understanding of what the determinants of the observed transitions in the phase patterns are.

For more information, you alsoshould read "Robot Control with Biological Cells" (PDF format, 15 pages, 262 KB).

So will we soon see biological cells becoming an integral part of robotic devices? Time will tell.

Sources: Various web sites

You'll find related stories by following the links below.