Research Paper #526
|Autonomous Robots: from Dream to Reality
|To be presented at Robotikdagarna, Tekniska Hogskolan, Linkoping, May, 1991
|This paper argues that the problems facing the constructors of autonomous robotic systems today are those of design rather than of performance. it examines one such problem, that of orchestrating the composite activity of a system in such a way as to achieve both robust, efficient performance and economy of design effort, and identifies four qualitatively distinct levels at which problems can be addressed in robotic systems. An example of what is possible given a suitable orchestration of effort is then discussed. In conclusion, it is suggested that the current situation, in which the trade-offs between different levels are obscure and the design rules for autonomous systems are largely empirical in intuitive (where they exist at all), is both dangerous and exciting: dangerous, since it is impossible to guarantee the performance of an autonomous robotic system in advance; and exciting, because the current state of affairs offers many interesting opportunities for uncovering the theoretical and practical laws which govern the organisation of competence in both robotic and natural systems.
|NO ONLINE COPY