Nervous Nets

Dateline: 12/01/98

Want to build a robot? You got what it takes to become successful robot builder (many hours of work, knowledge of mechanics, electronics and computer science), but simply cannot afford money needed to develop those deviously complex robot "brains". Don't worry: you can build your robot according to the philosophy of Los Alamos Laboratories researcher Mark W. Tilden, today known as BEAM robotics (an acronym for biology, electronics, aesthetics, and mechanics). It relies on the use of recycled parts - simple analog circuits as opposed to digital components and solar power instead of batteries.
The primary source of these components are broken walkmans and other electronics devices. You wouldn't expect "smart" robots can be built from such sources, but as it turns out they are capable of exhibiting surprisingly complex behavior. The most complicated part of constructing these critters is in the creativity of recycling junk parts into your design. Clever design results in BEAM robots that are autonomous, solar-powered and without "on/off" switch. They rely only on their built-in "instincts" to survive, and analog, not digital components - very few BEAM robots use microprocessors.

Some will recognize this kind of "minimalist" design in Brooks' subsumption architecture and other similar approaches. Other say BEAM is over-hyped, and that Tilden just renamed simple electronic circuits with complex sounding names. However, there is another detail that makes BEAM approach so unique: nervous nets.

This technology is based on the idea that a brain is no good without a spinal cord, so it acts very much like a silicon spinal cord. Robot without a brain? Look around, the majority of life species on this planet don't have a brain, and still function perfectly. As Tilden says, nervous nets are to the neural nets the same way peripheral spinal systems are to the brain.

Biomorphics (BIOlogical MORPHology) is basically the science of building machines that are "survivors" as opposed to "thinkers" or "workers". Biomorphs are not machines designed to perform a set of goal oriented tasks or work, but rather to express modes of survivalist behavior: the survival of a mobile autonomous machine in an a priori unknown and possibly hostile environment. No notion of conventional "intelligence" is used here, because intelligence in any sophisticated form is unnecessary for this concept. If we define life as that which moves for its own purposes, then we are dealing with living machines here.

Disregarding famous Asimov's laws of robotics, Tilden had its own rules that will guarantee an autonomous machine's survival. These Biomorphic Survival Laws are:

  1. A machine must protect its existence.
  2. A machine must acquire more energy than it exerts.
  3. A machine must exhibit (directed) motion.
They are very easy to incorporate into machine structure and control systems at a minimalist level. Complexity is reduced, and survivability enhanced.

As for the technical background, nervous (Nv) nets are by definition real-time non-linear analog control systems emulating a low-level peripheral spinal system. Based on arrays of sequential RC (Resistor-Capacitor)-time-based pulse delay circuits in closed loops, a Nv net is any circuitry that can act as a media for sustaining independent control "processes." The term Microcore generically refers to any Nv core greater than equal to four. These devices are "soft" designs, in that the environmental dimension must be absorbed, modified, and acted on for the devices to make successful headway through a complex world. These devices do not use "feedback" in the standard sense, but rather "implex", as the driving forces are augmented by perceived load rather than by a separate regulating path. Result: a fully adept insect-walker that is fully controlled and operated with as little as 12 standard transistor elements.

The biomorphs were already tested on the Yuma proving grounds, Arizona, that has been pelted with bombs, mines, missiles, and shells since the 1940s. They continued working even after multiple limbs and sensors have been removed or damaged, and to dynamically negotiate complex terrains as an emergent property of their operation.

Future work will concentrate on building light-seeking vision/head mechanisms powered by this technology. More recently Tilden has been doing some advanced work into the nature of minimal cognitive architectures. Future BEAM robots will definitely have better brains involving Nv arrays with hundreds of neurons. There are some surprising problems, for example how to keep robot and his sensors clean so that his performance remains constant. In the coming years, it is also hoped to be possible to demonstrate real machines to assess feasibility for nano-robotic applications.

More details on biomorph robots will be included in the Tilden's upcoming book Principa Robotica, along with a detailed explanation on what does what and why. This includes new hexatile boards, a Nv single sided square and hexagonal protoboard designed to take advantage of various symmetries in Nv circuitry, pre-arranged in sheets to provide the most convenient collections of biomech components for walkers. If you want to learn more about it right now, try the links below.

Similar site(s) of interest: