Morphological Computation

Leverhulme Trust Project: Computing with spiders’ webs – An inspiration for new sensors and robots

This is a 3 year Leverhulme Trust Research Grant project by Helmut Hauser (PI, University of Bristol) and partner Fritz Vollrath (Co-PI, Oxford University). 

Abstract

Spiders use their highly dynamic and complex webs not only as simple traps, but also as sophisticated signal processing devices that help them to categorise precisely and robustly vibration patterns introduced by prey, mates and other events. This project combines biological experimentation, simulation and mathematical modelling to understand and extract the underlying principles, and to infer and build disruptive designs for novel flow and vibration sensor technologies. In addition, inspired by the capability of spiders to build their own signal processors, we will develop robotic prototypes capable of deploying our novel sensor technology on demand.

Summary

The spider's web is a sophisticated structure created by a dedicated and complex behaviour pattern optimised through evolution over approximately 100 million years to serve the ultimate purpose of catching prey [1]. The efficacy of the web as a trap depends heavily on the correct and robust categorisation and localisation of various events, including trapped prey, potential mates, broken threads, wind, and others. In particular, the correct identification of prey, e.g. whether wasp or fly, i.e. dangerous or not, is crucial for their survival.

Spiders are able to carry out this complex computational task robustly using only mechanoreceptors on their legs that allow them to measure vibrations. Some species are even able to do this by monitoring only a single thread. These abilities suggest that the web is not only a static, passive structure, but rather actively contributes to the pattern recognition task.

Further support of this point of view is provided by the observation that the components of the employed building materials, although they are all basically (biochemical) silks of one type or other, have hugely different physical properties depending on their location within the web. This results in a much more complex and dynamic structure [2],[3],[4] than one would expect from a simple, static trap.

We suggest that the spider’s web is very highly likely to be a nonlinear, dynamic filter, i.e. a pre-processing unit that helps the animal to interpret vibration signals. The web’s dynamic properties and its complex morphological structure naturally enables the mixing of vibration signals in a nonlinear fashion at their nodes, damping of unwanted frequencies, and even exhibiting memory by transmitting and echoing the signals throughout the network [5]. In effect the spider devolves critical computational load to its web. The concept of outsourcing computation to a physical body (e.g., from the brain to another part of the body) is usually referred to as morphological computation [6],[7],[8],[9] and can be observed in all biological systems [10].

Taking all this into account the spider’s web can be perceived as a highly dynamic, morphological computation device and, moreover, an externalised computational resource that the spider is able to build on demand. To understand the underlying principles of these mechanisms this project will systematically investigate and quantify various aspects of computation (e.g. filtering capabilities, memory, signal integration, etc.) in biological webs of various species.

Besides this new understanding of the biological system, the project goes a significant step further. The idea of intelligent morphological structures is not just useful for spiders, but can be exploited to develop novel, intelligent sensor technologies, especially, in the context of vibration and flow sensors.
In addition, the idea of on-demand deployable morphology-based computational devices is of great interest for robotics. Potential applications include maintenance robots that swarm through tubes building flow sensors at key points to detect anomalies, or climbing robots that construct vibration sensors at strategic places on buildings to detect earthquakes or structural failure.

Objectives

In summary, the objectives of his project are:

• To understand the underlying computational aspects of spiders’ webs

• To obtain fundamental design principles for intelligent morphologies capable of carrying out useful computations through vibration

• To design and build novel vibration and flow sensors based on these principles

• To design and build robotic prototypes capable of deploying these sensors on demand

References

• [1] Vollrath, F. & Selden, P. (2007) The Role of Behavior in the Evolution of Spiders, Silks, and Webs. Annual Review of Ecology, Evolution, and Systematics, 38: 819-46
• [2] Emile, O. Floch, A.L. & Vollrath, F. (2006) Shape memory in spider draglines. Nature, 440 (7084): 621
• [3] Lin, L., Edmonds D. & Vollrath F. (1995) Structural engineering of a spider's web. Nature, 373 (6510): 146-148 [4] Schneider, J. & Vollrath, F. (1998) The effect of prey type on the geometry of the capture web of Araneus diadematus. Naturwissenschaften, 85 (8): 391-394
• [5] Mortimer, B., Gordon, S.D., Holland, C., Siviour, C.R., Vollrath, F., & Windmill, J.F. (2014) The speed of sound in silk: Linking material performance to biological function. Advanced Materials, 26 (30): 5179-5183
• [6] Hauser, H., Ijspeert, A., Füchslin, R., Pfeifer, R. & Maass, W. (2011) Towards a theoretical foundation for morphological computation with compliant bodies. Biological Cybernetics, Springer Berlin / Heidelberg, 105, 355- 370
• [7] Hauser, H., Ijspeert, A., Füchslin, R., Pfeifer, R. & Maass, W. (2012) The role of feedback in morphological computation with compliant bodies. Biological Cybernetics, Springer Berlin / Heidelberg, 106, 595-613
• [8] Nakajima, K., Hauser H., Li T.; and Pfeifer R. (2015) Information processing via physical soft body. Scientific Reports 5, Article number: 10487, doi:10.1038/srep10487
• [9] Nakajima K., Li T., Hauser H., and Pfeifer R. (2014) Exploiting short-term memory in soft body dynamics as a computational resource. Journal Royal Society Interface, 6 November, vol. 11, no. 100, 20140437, DOI: 10.1098/ rsif.2014.0437
• [10] Pfeifer, R., & Bongard, J. C. (2006). How the Body Shapes the Way we Think. The MIT Press.