Insectile and Vermiform Exploratory Robots

NASA's Jet Propulsion Laboratory, Pasadena, California

A six-legged robot resembling an insect and a legless segmented robot resembling a worm (see figure) have been proposed as prototypes of biomorphic explorers small, mobile, exploratory robots that would be equipped with microsensors and would feature animallike adaptability and mobility. Biomorphic explorers and related concepts have been described in several previous articles in NASA Tech Briefs, the most relevant being "Biomorphic Explorers" (NPO-20142), Vol. 22, No. 9, (September 1998), page 71 and "Earthwormlike Exploratory Robots" (NPO-20266), Vol. 22, No. 6, (June 1998), page 11b.

Depending on the specific environment to be explored, a biomorphic explorer might be designed to crawl, hop, slither, burrow, swim, or fly. Biomorphic explorers could be used for such diverse purposes as scientific exploration of volcanoes, law-enforcement surveillance, or microsurgery. Another potential use for biomorphic explorers is detection of antipersonnel mines; there is a pressing need for robots that could be deployed in large numbers to detect antipersonnel mines left on and in the ground after armed conflicts. The proposed six-legged robot would be designed with a view toward that application. There is also a need for burrowing robots that could search earthquake rubble for survivors; the proposed vermiform robot would be suitable for this purpose.

Robots That Look and Move Like Small Animals would be developed for use in a variety of exploratory tasks. Six-legged robots could be developed into a mass-producible, mass-deployable units to search for antipersonnel mines. Legless robots similar to the one depicted here could burrow in earthquake rubble to search for survivors.

The proposed six-legged robot would be capable of traversing various types of terrain. The legs would be attached to a main body at shoulder ball pivots. Rotations at the shoulders would result in translations of the feet. The legs would feature telescoping segments that could be lengthened or shortened to suit the direction of motion and the terrain. For example, the legs could be shortened to obtain greater mechanical advantage for climbing, or lengthened to increase speed in level or downhill travel over smooth terrain. The legs would be tipped with footpads that could be configured to suit the terrain. For example, a scissorlike arrangement of footpad members would be use on hard terrain (e.g., rocks), while the footpad members would be spread out to form a larger contact area on soft terrain (e.g., sand). The legs and footpads would be actuated by springs paired with shape-memory-alloy (SMA) wires; within each actuator, the spring would pull or push in one direction, while the SMA wire would pull in the opposite direction by an amount that would be changed momentarily by passing a momentary electric current through the wire to heat it momentarily above its shape-memory transition temperature.

The proposed vermiform robot would be capable of both anchored rectilinear motion similar to peristalsis and a transverse motion, based on the motions of Amphisbaenia - a legless order of reptiles that burrow with notable efficiency. The anchored rectilinear motion would be effected by anchor modules that would look like cones paired base to base. Within each anchor module there would be a pistonlike assembly actuated by pairs of springs and SMA wires. The assembly could be actuated to either (1) shorten the module longitudinally and expand the outer cone radially to anchor in the wall of the burrow or (2) lengthen the module longitudinally and retract the outer cone from contact with the tunnel wall. For example, suppose that all anchor modules were initially in the minimum-diameter, maximum-longitudinal-length configuration. The foremost module could be expanded radially to anchor the head end, then the next module could be expanded, and so forth, in sequence from front to rear. The longitudinal shortening accompanying the radial expansion of each module would draw the trailing modules forward.

The anchor modules would be connected by collars of a flexible material in which SMA wires would be embedded at multiple circumferential positions. The SMA wires would be oriented longitudinally. The wires could be energized selectively to bend the collar; in this way, part or all of the robot body could be arched.

In both robots, artificial neural networks would receive inputs from sensors and would respond by issuing commands for the SMA actuators to effect complex combinations of motions to achieve the overall lifelike mobility. Artificial neural networks were chosen for this application because they appear to offer the maximum potential for achieving a desired combination of capability for learning, adaptability, fault tolerance, composability (ability to smoothly integrate various primitive motions into complex motions and other activities), and generality to enable application to future biomorphic explorers.

This work was done by Sarita Thakoor, Brett Kennedy, and Anil Thakoor of Caltech for NASA's Jet Propulsion Laboratory. NPO-20381