Posts Tagged ‘Maze solving robot’

1971 – Model 2004 Maze-Solving Computer – Richard Browne (American)


Source: Xenia Daily Gazette Mon, May 24, 1971

Computerized mouse maze first of 3 long-term projects for Xenian.

by Ward Pimley – Gazette staff writer

To a research psychologist, running a mouse through a maze to investigate behavior patterns is a common occurrence. But to an electronic engineering drawing specialist who wants to simulate the test, various alterations must be made.
Richard Browne, a drawing specialist in his seventh year with Systems Research Laboratories, Inc. (SRL), has recently finished a lengthy project designed to propel a wooden mouse through a maze with directions being supplied by a computer. He resides at 2004 Tahoe Dr.
COMMONLY, referred to as cybernetics, the system constructed by Browne uses a computer attachment which receives data from the mouse as to its location and the presence or absence of barriers, The computer then tells the mouse which direction to move based upon the data. While the mouse is searching for its "cheese," a metal block which short circuits the electric charge upon contact, the computer is storing in its "memory" information pertaining to the maze; that is, where the alley blocks are and what routes are beneficial to the mouse's search for the goal.
CYBERNETICS is a branch of science which mechanically and electronically attempts to reproduce the human thinking process into machines. Browne's computer, designed to comply with this principle, is programmed to receive information from the mouse, "analyze" the situation, then direct the mouse on its journey. The mouse, one-inch creature carved from balsa wood, has two copper whiskers which signal the computer when the mouse has bumped into a maze barrier.
Directional information is then sent back to the mouse whereupon an electromagnet beneath the aluminum maze moves the mouse in the direction indicated. The electromagnet is driven by two one tenth horsepower engines which control both north-south and east-west movements of the mouse.
THE COMPUTER, 600 pounds of wires and relays, has the capability of processing both partial and total accumulations of "knowledge." The partial knowledge refers to the store of information regarding the squares in which the mouse has investigated, while the total accumulation is the computer's memory of the correct path the mouse should take to solve the maze. After the mouse has found the goal, it may be placed anywhere along the proper path and it will move directly to the goal without either making detours or bumping into alley walls. There is an exploration strategy which the mouse follows, Browne explained, every time it enters a square. Five steps are involved, all occurring within one-tenth of a second. The procedure is repetitive and designed so that the mouse will examine all possible avenues of escape from a square. If the mouse should encounter a wall in one direction, it then turns 90 degrees clockwise. If there is no wall in that direction, the mouse will exit the square. Otherwise, it will turn again to check a new direction. There are 25 squares on the maze with removable walls for reshaping the maze. Browne, said there are 873 duodecillion/(873 followed by 12 zeroes) solvable maze patterns possible in his operation. Should the mouse solve one million maze patterns per second (clearly an impossible task), it would take the mouse 2.7 septillion (seven zeros) centuries to solve all possibilities, Browne said.
THE PROJECT took Browne 10 years to complete, working on and off, he said. The idea for the maze came from a May 1955 issue of Popular Mechanics where an article was printed about a man who had completed such a project. Browne decided to duplicate the feat, although he designed and constructed the computer by himself. While Browne built his unit with spare parts, he said that a computer and maze constructed from new parts (and including labor costs) would cost about $15,000. The present project is completed, Browne said, except for a couple of minor improvements to be made. One of these is to put wheels on the mouse to facilitate easier movements. The other is to replace the magnet in the mouse with a stronger one so that the mouse will not escape from the electromagnet's pull from under the table.

However, Browne is not quitting his dabbling with home made electronic projects. He presently has in mind two further projects to operate from the computer he has already built. One of these is a model railroad, which Browne estimates will take him 15 years to complete (working on and off of course). The other is an electromagnetic calculator which will perform complicated mathematics.

Man's creative urge, it seems, still lives in Richard Browne.

See other early Maze Solving Machines & Robots here.


Tags: , , , , ,

1954 – Maze-Solving Machine – J. A. Deutsch (British)

Anthony Deutsch, aged 26, with his maze runner from Oxford University's Institute of  Experimental Psychology.

The head-lamp on the trolley is turned on, and various photo-electric cells are mounted at strategic points in the maze.





Deutsch's Maze runner was considered the most sophisticated at the time. It was capable of transferring its training from one maze to another which is topologically equivalent even though its lengths and shapes have been altered. Deutsch's maze-runner also takes advantage of short cuts added to the maze. (from Martin Gardner 1961)

The system consisted of six uniselectors and twenty-three relays. The uniselectors have six wipers each, and are ordinary Post Office (P.O.) equipment. Only three positions are employed on each. The relays are similarly large P.O. relays with four "make" and one "break" contact.

Deutsch, J.A. "The Insightful Learning Machine", Discovery 16:12 1955 pp 514-517. pdf  here

Deutsch, J.A. "A Machine with Insight", Quarterly Journal of Experimental Psychology Vol.6 part I pp 6-11.  pdf here.

See other early Maze Solving Machines & Robots here.

Tags: , , , , , ,

1977 – “Maizie the Robot” – David Strange (British)

Maisie the Robot (c1977) with David Strange.

From a comment below, David Strange informs me that the robot he called Maizie was a maze solving robot whilst he was working for the BBC Open University. It was used in several programmes on cognitive psychology. Thanks David.

From further correspondence with David (Dec 2015):

The whole of Maizie was scratch-built using discrete components.  In those days I got 2 to 3 weeks notice for an OU programme requirement and just got on with it.  There was a day for conceptualising and planning, then off I went and bought the bits I thought I was going to need and use.

I bought the components, cut the metal, designed and built servos, integrated them with the electronics, figured out how to get reliably get exactly 45 deg turns, detect the end of a row and just got on and designed the logic.  

I even scratch-designed electronics for a 2 or 3 word speech recognition system so the programme presenter could ask Maizie to set off and solve the maize.  I recorded the presenter saying the phrase he intended using and took the tape to the BBC research department, who did a frequency analysis for me against the timing of the phrase.  I then used the data to design a circuit that generated the start pulse for Maizie, thanks to the integration of phase-lock-loop chips that were just available with 555 timers and some bits of general electronics and logic chips.  It was just case of using a set of narrow band pass filters and looking for the right frequencies in the right order at the right times.  It worked. … microprocessor(s) were not used.

See other early Maze Solving Machines & Robots here.