William Grey Walter

INT D 554/PSYCO 457 • William Grey Walter built the first Week 7: Machina Speculatrix autonomous robots in 1948 and 1949. • They were recently rediscovered and recreated in Bristol The Tortoise Reviewed • His work is described in the 1953 book The Living Brain Our LEGO Tortoise • He was a “scientific character” – “His popular and academic reputation encompassed a heterogeneous series of roles ranging from robotics pioneer, home guard explosive experts, wife swapper, t.v.- pundit, experimental drugs user, and skin- diver to anarcho-syndicalist champion of leucotomy and electro-convulsive therapy.” (p. 616).

Machina Speculatrix A Simple Machine

• We are interested in Grey Walter because of his • Grey Walter’s research program “held promise of demonstrating, or pioneering work in building autonomous robots, at least testing the validity of, the the famous Tortoise or Machina speculatrix theory that multiplicity of units is – “An electro-mechanical creature which behaves so not so much responsible for the much like an animal that it has been known to drive a elaboration of cerebral functions, not usually timid lady upstairs to lock herself in her as the richness of their bedroom, an interesting blend of magic and science” interconnection” • In this sense it was a reaction against Ashby’s homeostat • Here is some classic footage of the robot in action • And some more footage…

Recording Complex Behavior Elsie’s Behavior: Getting Around?

• Candles were fixed to the turtles’ • “The machine shells, and long exposure photographs were taken as the circles around [the machines moved light] in a complex • Traces of light in the photographs path of advance recorded the paths of the machines andithdd withdrawal” – “Some of these patterns of performance were calculable, though only as types of behaviour, in advance; some were quite unforeseen” (Grey Walter, 1953, p. 130)

1 Elsie Returns Home Elsie’s Complex Behaviour

• “Started in the dark • A screen hides the the creature finds its light, it can’t be seen way into a beam of light and homes on • Elsie oscillates, but the beam into its then hits and moves feeding hutch” the screen • Why does she do • When the light this? appears, she circles it • Is this different than at a distance the prior behaviour? • Why?

Elsie’s Behaviour: Avoidance Elsie’s Behaviour: Free Will

• Elsie successfully • “The solution of the dilemma of Buridan’s ass. The photoelectric cell which functions avoids a stool and as the creature’s eye scans the horizon continuously until a light signal is picked approaches the light up; the scanning stops, and the creature is • What must be built directed towards the goal. This mechanism converts a spatial situation into a temporal into Else to permit one and in this process the dilemma of two symmetrical attractions is automatically this to happen? solved, so that by the scholastic definition the creature appears endowed with `free- will’. It approaches and investigates first one goal and then abandons this to investigate the other one, circling between the two until some other stimulus appears or it perishes for want of nourishment.”

The Parable Of The Ant Elsie’s Social Behaviour

• “Viewed as a geometric figure, • Elsie dances with the ant’s path is irregular, Elmer, and vice complex, hard to describe. But versa, because of its complexity is really a the candle complexity in the surface of the beach, not a complexity in the • The two then race ant” (Simon, 1996, p. 51) to the hutch, with • One way to complicate Elsie’s Elsie winning environment is to include • Where does this another agent behaviour come from?

2 Basic Principles The Next Step

• Two radio tubes • The behaviour of the • Two sensors (light and touch) tortoises was very complex • Photocell is linked to steering mechanism • One reason for this • No light -- continuous forward complexity was the fact exploration combined with that theyyy were “many- oscillation situated” – sensed more • Weak light -- steering inhibited, and machine moves than one aspect of the forward towards the light environment • Strong light -- dazzled state, • Our goal is to build, steering kicks back in at double speed program, and observe a • Touch sensor activates more similar robot ourselves! random oscillation to permit obstacle avoidance

3