MazeBot

Theseus the Mouse, 1951

In characteristically curious pursuit, Claude Shannon puts 90 telephone switching relays toward a new task. Theseus the mouse would indeed navigate his way into the legends of early computing, becoming widely regarded as one of the first examples of machine learning. The mouse (the maze) would remember and avoid walls encountered, even if moved arbitrarily.

Theseus the Mouse, 1951

Claude Shannon shows off the 5x5 magnetic reed switch array, 90 relay switches and x,y gantry mounted magnet under the hood of his maze.

MazeBot

Slotting maze stantions.

MazeBot

Flycutting stantion tops.

MazeBot

Fit test mock-up (top left) and getting an early look with metal components.

MazeBot

Probe version 2. Mag release with extension spring replacing compression, which reduces oscillation in probe tip.

MazeBot

X axis rail replacement project. Substantial reduction in mechanical slop, MGN12 linear rail replaces LM10UU style.

MazeBot

Showing control panel fitted out, featuring push button switches, on-off toggle, and RGB LED indicators.

MazeBot

Showing cable routing and CW from left: mounted Pi zero, 5v voltage controller board with display, Arduino and CNC hat, auxiliary breakout board.

MazeBot

Control cabinet open for inspection.

PlotBot

This program takes electrical input from the maze (walls, goal) and decides where to go next, or if the maze even has a solution!

PlotBot

This program manager starts on boot up, and starts or stops all other programs. It takes input from the control panel. Even simple tasks, e.g. blinking an indicator light requires calling a specific subprocess!