1965 – ADROIT Simulated Insect – Leonard Friedman (American)
At this stage I have very little information on Friedman's ADROIT. If I can get hold of his paper I will add exceprts here.
interesting reference from SRI Shakey's 1966 interim paper
"Definition of the Automaton System
Factors Influencing the Choice of a Mission
There is not much prior art in the area of automata controlled by a modern , stored-program computer. Devices like Grey Walter ' s turtles, Claude Shannon's maze-solving rat, and the Johns Hopkins' beast are clearly ambulatory automata , but their motor functions are controlled by very rudimentary logical devices that do not qualify as computers. Space probes like the Surveyor series respond to instructions formulated through the use of general-purpose machines , but (except in flight) they are neither mobile nor autonomous. Ernst4 has reported on the use of a computer to control a general-purpose "hand" ; however limitations in both the electromechanical system and the computer used restricted the capabilities of this device. Friedman5 has programmed a simulated behavioral model of an insect , ADROIT , that locates and transports
certain objects in its environment. In the latter work there was no actual vehicle–the simulation was in the form of a CRT scope presentation.
The examples cited above have the common feature of being rather specialized toward solving one particular problem or illustrating one particular mode of behavior. An extreme example along similar lines
would be a chess-playing automaton that could do nothing but manipulate chess pieces under the control of a program designed exclusively for that purpose. In selecting a mission for Mark I we wanted to avoid (if possible) tendencies to over-specialize in any of the areas consisting of funct ional modes of behavior , tasks to perform , or task-oriented problem solving. Another point we wished to make sure of was that interaction of the automaton with its environment would be essential to the execution of tasks making up the mission. That is, the environment (or world) of the vehicle would not merely be a convenient arena for the display of virtuosity in some skill , i. e . , an unusual form of locomotion or, perhaps broom balancing. One more philosophical point–we decided to make no attempts to create a system that models something else so that we might then study the model. For example , we would refrain from trying to simulate things like animal behavior or human cognition unless it was expedient (and possible) to do so as part of a tactic for carrying out some useful task."
Friedman has programmed a simulated behavioral model of an insect, ADROIT, that locates and transports certain objects in its environment. In the latter work there was no actual vehicle–the simulation was in the form of a CRT scope presentation.
L. Friedman, "Theories of Instinctive Behavior and their Computer Analogues " Document No. SP-2292/000/00 , System Development Corp. Santa Monica , California (10 December 1965).
Handle / proxy Url: No Full Text PDF Available
230300 – Bionics
050800 – Psychology
Theories of instinctive behavior proposed by Lorentz (Physiological mechanisms in animal behaviour. Symp. Soc. Exper. Biol., 1950, v.4, p221-268) and Tinbergen (The study of instinct. Oxford, Clarendon Press, 1951) have been analyzed with respect to longstanding criticisms leveled at them, as well as to neurophysiological evidence. A theory of instinct that meets the criticisms and reconciles the ethological and neurophysiological evidence is proposed. This theory retains the fruitful concept of the Innate Releasing Mechanism (IRM) introduced by Lorenz and unifies the description of instinctive behavior in terms of hierarchies of feedback loops organized as Behavior Units. The Behavior Units are activated by a complex organization of IRM's and Selection Units operating continuously among a host of competing pressures, internal and external, to select the most appropriate action. Various assumptions about the role of learning in instinct-dominated animals are exposed and examples are given of the connection between our assumptions about learning and the modified theory of instinct. To validate the theory and permit its development in greater detail, a computer simulation of the theory has been undertaken, using an artificial animal, ADROIT, as the 'guinea pig'. Progress to date with this simulation is discussed. (Author)