Artificial Neural Systems for Robots

History

Despite the construction of many ingenious mechanical automata over the centuries (including chess playing Turks and flatulent ducks (Wood 2003)), it was not until the 1930s that devices recognizable as robots (in the present day sense of the term) appeared. Early mobile robots, such as Thomas Ross’s ‘robot rat’, completed in 1935, were designed for narrowly focused single behaviours (often maze running) and employed highly specific mechanisms to achieve their intended task (Cordeschi, 2002). These ‘artificial animals’ inspired what were probably the very first examples of more general mobile autonomous robots – W. Grey Walter’s tortoises (Walter, 1950). These robots were also the first to employ an early form of neural network as their artificial nervous system. They were born out of the cybernetics movement, a highly interdisciplinary endeavour – drawing together pioneers of computing and modern neuroscience – which was the forerunner of much of contemporary AI and robotics, and the origin of artificial neural networks and evolutionary computing, as well as control and information theory (Boden, 2006; Husbands et al., 2008).