Hyper-interspersed nano/MEMS-architecture design for new concepts in miniature robotics for space exploration

Launch weight and volume requirements are substantially decreased by reduction of probe size in exploration mission systems, as mass and volume both scale as the third power of system size. Accordingly, the already quite developed ME MS (MicroElectroMechanical System) technology, that offers low cost, small, fight weight, and increasingly reliable devices through durability and redundancy, is strongly attractive as a near-term technology for significantly reducing the cost to launch and operate space systems. It is shown that the final goal of MEMS technology, i.e. the merging through solid state microdevices of the functions of sensing, computation, communication and actuation, can lead to a new, biomimetic kind of miniature robotics, particularly suitable for planetary exploration, through molecular mono-electronics/MEMS integration jointly with a hyper-interspersed architecture made up of autonomous units embodying sensors, information processors and actuators. The problem tackled here concerns the basic design of such miniature robots, from some mu m to insect size, featuring finely structured intelligent autonomous parts as smart skins, sensory and manipulating members working on the analogue external reality and communicating with their inner molecular level nondiscrete pseudo-analogue information processing networks. The (mesoscopic network)/MEMS units are shown to embody a quantum mechanical/macroscopic world connection, in which the nondiscrete molecular devices allow the automaton parts to perform very complex, Fast information processing operations as metaphores of bionic Functions like learning, attention, and decision making under uncertain conditions, this last due to the stochasticity inherent in the quantum network. Flexible architectures instead of von Neumann type rigid architectures in addition to hyper-interspersion of autonomous units can be realized through such nano/MEMS devices, and the mu m - cm size of the whole robots and their organs allow dynamic biochemical, possibly reaction diffusion, spatially separated highly nonlinear systems to be exploited as additional primitive computing devices (e.g. chemical oscillators, dissipative biomolecular distributed systems, planar photoactivated enzyme biosensors). Each interspersed unit can be designed as a multilevel nondiscrete system according to the information processing "rank" to be obtained in simulating the biological nervous system activity. (C) 1999 Published by Elsevier Science Ltd. All rights reserved.