- Article in Robotics Age Jan/Feb 1982.
|Quester by David Buckley
Article in Robotics Age Jan/Feb 1982.
- Quester - Retired
- A micromouse, which uses vision to detect the walls of
the maze, segmented bump sensors in case it
doesn't, and a layered behaviour control program.
Finalist in the Robotics Age "Home Robot Photo
Contest". Article in
Robotics Age Jan/Feb 1982.
Photograph of Quester used to illustrate an
article on mobile robots, p641, The Home Computer
Advanced Course #33, 1984.
Photograph - Scale Models International, April
Built June 1981 -.
Size - 8" * 5 1/2" * 7 1/2".
Operational area - 3ft * 3ft using a small maze.
# Technical overview
1981 - Quester took part in the Wembley heats of the July 1981 Amazing Micromouse Contest
1981 - Paris, September 7-10, Quester took part in the EUROMICRO competition
1982 - David L Buckley won third prize in the schools competition of the Euromouse Maze Contest
A £25 computer book voucher from McGraw Hill
1983 - David L Buckley's mouse Quester won 4th prize in the 1983 finals at Earles Court, Novice Section
1996 - Quester won the 'Thezeus Award 1996' created and presented by Alan Dibley.
[The maze for a micromouse has white wall (with red tops) and a black baseboard.]
Quester has a Reactive subsystem ( Before Brookes, Arkin, Tilden et al. ) and a top level control computer.
Quester still works but the tape player for loading programs doesn't work and neither do the program tapes.
- Quester has a differentially steered base with quadrature shaft encoders in each wheel gear train.
- Light from subminiature LEDs (the size of SMT ones) shines through holes cut in the gearwheels and detected by phototransistors.
- The phototransistors feed into hardware logic quadrature decoder counters which are read by the control computer.
- There is a forward facing Ultrasonic Tx/Rx sensor pair activated and read by the control computer.
- There are four wire loop bump detectors (a bit like Shakey) which feed into a CMOS logic Reactive subsystem.
- the Reactive subsystem uses resistor/capacitor delays to sequence commands to the motor drivers for avoidance action.
( As Mark Tilden used later for Beam robots. )
- There are two compound eyes built with phototransistors, each with three auto iris pixels.
- The acceptance angle of each pixel is about 5 deg.
- The auto irises have about 4 times the acceptance angle of the associated pixel.
- One pixel of each eye is set to detect side openings to a maze corridor.
- One pixel of each eye, looking slightly down, is set to detect a wall in front.
- The front pixel can be used to detect the distance from the wall by sensing the transition from black to white as the wall is approached.
- One pixel of each eye is used for guidance and is aimed out and down at the baseboard just by the sidewall about 6 inches in front of the robot.
- detecting the light dark transition allows Quester to be steered down a maze corridor.
- Using the front facing pixel or the guidance pixel of each eye Quester can square up to a wall.
- The Brain is a 6502 control board with a 512 byte Monitor, 1K RAM, a LED readout, hex keypad and tape interface and cost today's equivalent of a high-end desktop PC.
- The computer was programmed in hex from hand assembled code and programs saved to tape.
- The computer has direct control of the motors and Ultrasonic sensor.
- The computer can read the eye receptors and the shaft encoder counters.
- Using the shaft encoder counts and a PI error loop (no D) turns can be made to about 6 degrees which is enough for orientation in the maze.
- Using the shaft encoder counts and a PI error loop (no D) distance can be measured to about 1/4 inch which is enough to count maze cells.
- Using the information from the guidance pixels and a PI error loop (no D) the robot can steer down a maze corridor.
- The control computer can read and override the hardware Reactive subsystem. ( very important! )
Hand assembly takes forever especially as links need to be altered after an edit. (Quester's computer was my ONLY computer).
The 1K RAM was too small for Quester to be more than a micromouse, even a simple maze map took 256 bytes.
I did fit the arm and gripper but realistically there was not enough code space and decoding what the eyes saw was too hard.
- Scale Models International, Argus Press, Hemel Hempstead.
- The Home Computer Advanced Course, Orbis Publishing Ltd, London
(Published in Europe, South Africa, New Zealand, and Australia).
See also Micromouse - the early years