Tomorrow’s super robots may owe their mobility to a cockroach’s legs today

The marriage of machine and biology requires adopting the pliability and strength from the legs of this despised insect

The cockroach is an insect despised for its ubiquitousness, among other reasons. Yet, it may hold a key to the next evolutionary step in the “life” of robots.
Background

For years, serious futurists could only imagine that robots, such as the television model, would always be stiff, clumsy, and prone to breakdown. This was before the advent of “Biomimetics,” a research aimed at developing a new class of biologically inspired robots that exhibit much greater robustness in performance in unstructured environments than today’s robots.

This new class of robots will be substantially more compliant and stable than current robots, and will take advantage of new developments in materials, fabrication technologies, sensors and actuators. Materials found in nature differ significantly from those found in human-made devices. Nature appears to design for “bending without breaking” and employs tissues that are compliant and viscoelastic rather than stiff, homogeneous, and isotropic. In addition, local variations in biological materials, tailored to meet local variations in loading, are common. The nonlinear, compliant, and inhomogeneous materials found in even the simplest animals provide them with a sophistication and robustness that today’s robots cannot match. And it is hard to find an animal as simple as the cockroach.

Actually, the deathhead cockroach possesses legs with compliant muscles and skeletal components that increase dynamic stability and disturbance rejection. As the ability to analyze and fabricate mechanisms with compliant and functionally-graded materials improves, the opportunity exists to develop robots whose structures draw inspiration from simple animals such as insects and crustaceans. One fertile area for biomimetic design is the leg of walking or hopping robots, where leg compliance is especially important.

One method for manufacturing such robots is Shape Deposition Manufacturing (SDM), a rapid prototyping technology. SDM addresses many limitations of traditional manufacturing and assembly by enabling the in situ fabrication of mechanisms with complex geometry and heterogeneous materials. Design and fabrication of layered and heterogeneous materials (also called Functionally Graded Materials – FGMs) has recently been a focus of research. FGMs enable control of local variations of biomimetic components by selectively depositing soft and hard materials. To produce biologically inspired components of biomimetic/mechanical properties, a bridge between biological findings and SDM design specifications was required.

The first demand for SDM is to characterize biological structures and translate the characteristics into quantitative specifications for mobile robots. The second requirement is to model SDM material behavior to facilitate component design to meet these specifications. To address these requirements experiments were conducted on a hind leg of Blaberus discoidalis and described its response to both step displacement inputs and sinusoidal displacement excitations. Next, a test was carried out on one of the materials used in SDM, a soft polyurethane polymer largely used as joint material in manufacture, and fit the results to standard viscoelastic (pliable yet sturdy) materials and models. Comparison and understanding of the mapping between these two studies enable us to begin to design and manufacture legs similar to those found in biology.

The Study

The authors of “Material Modeling for Shape Deposition Manufacturing of Biomimetic Components,” are Xiaorong Xu, Wendy Cheng, Mark R. Cutkosky and Motohide Hatanaka from Stanford University, and Daniel Dudek and Robert J. Full at the University of California at Berkley, Department of Integrative Biology, Berkeley, CA. The authors are presenting their work at “The Power of Comparative Physiology: Evolution, Integration and Application” meeting, sponsored by the American Physiological Society (APS) August 24-28, 2002 at the Town & Country Hotel, San Diego, CA. To learn more about the conference and presentations, go to: http://www.the-aps.org/meetings/aps/san_diego/home.htm

Methodology

Relaxation and dynamic experiments were carried out on the hind leg of Blaberus discoidalis to aid in the selection of a material behavior model and to quantify measures of roach leg response. During testing, the coxa of the ablated metathoracic limb (hind limb) of the cockroach was epoxied to 3/8″ acrylic such that the coxa-femur and femur-tibia joints were free to rotate. Cyanoacrylate was used to attach one end of a stainless steel pin to the distal tip of the tibia; dental impression compound was used to adhere the other end of the pin to the arm of a servo-motor system. The leg was then displaced with the Aurora system, which is based upon a high performance rotary moving coil motor supported by precision ball bearings. The results are that the total error in the force-displacement measurements to be less than four percent that of a viscoelastic solid.

Results

The results indicate that a cockroach leg excited in a direction orthogonal to the joint direction behaves similarly to a viscoelastic material. The exponential nature of the force relaxation curves suggests viscoelasticity. The hysteretic nature of the force-displacement curves indicates that there is energy loss due to the internal friction, which is a common characteristic for viscoelastic materials. The cockroach leg is subject to a combination of bending and torsion in the experiment. The overall effect can be modeled as a torsion spring with a moment arm. Additional assumptions for the model include: (1) the axis of rotation for the leg is constant during torsion and (2) the joint material can be approximated using a lumped-parameter element with uniformly distributed linear viscoelastic properties.

The SDM process allowed an integration of a range of desired impedance into the structure of robot legs for improved robustness and simpler control. SDM-compatible materials span a wide range of material properties and the SDM process enables researchers to control local variations through Functionally Graded Materials (FGM). With information regarding the mechanical behavior of animal legs and the material characteristics of SDM materials, the researchers developed guidelines for biomimetic leg design.

Conclusions

Some polymer materials that can be used in SDM are similar to the biological materials found in insect legs that exhibit viscoelasticity. This inspires us to develop material models and design methodologies that can be used to guide biomimetic robot leg design and material selection. In this paper, we have discussed a simple linear, lumped parameter model used to characterize cockroach leg behavior in relaxation experiments and in response to sinusoidal excitations. We have also developed a dynamic test machine and begun characterizing a polyurethane material used for SDM fabrication of robot joints.

The current models of leg response assume linear viscoelasticity. The correlation between these models and the results of the experiments is relatively good at low frequencies and small displacements, but deteriorates at higher frequencies and displacements as nonlinear effects grow pronounced.

In addition, at very low frequencies, dynamic tests on cockroach legs indicate a higher loss modulus than that predicted by a standard linear model. Should these nonlinear aspects of leg behavior prove important for locomotion, the researchers believed that better models had to be developed better models to simulate the viscoelastic behavior of the leg in a wide frequency range.

Additionally, to produce legs with mechanical response similar to that of the real cockroach leg, enhanced characterization of additional SDM materials is required. Knowledge of SDM material behavior, along with information about the aspects of leg behavior important to locomotion, will enable the issuance of general design guidelines for designing biomimetic legs.

(It is worth noting that these legs have been used to produce a remarkable successful robot from Stanford named SPRAWL. SPRAWL can negotiate rough terrain without a brain or any reflexes because the control is built into the smart or tuned legs described above.)