Nearly Talking of Robots

My colour recognition machine

My system, my picture.

Whilst I don’t have a great deal of time to devote to them nowadays my robot and train projects continue (slowly) as my occasional escape from care giving. Last night, I became so frustrated with my inability include a short video clip in a reply to a note (most likely operator error), that I decided to create this new post.

It started with a discussion of AI. I was overcome by this sudden urge to create a machine that was primitive, but which could nonetheless learn and recognise colours. So that’s what I did. And here it is.

OK, it is not *exactly* a robot. Nor is it especially useful, but I love it all the same. It is a simple functioning sensory system that “sees” and reports colour.

Sound right up…🔊

Loading video

The HC-SR04 ultra sound detects the target. The command Mega fires a relay to flood the target with light from 2 rows of LED strips, the sort used under kitchen cabinets.

The target cards are a different colour on each side.

The colour sensor consists of a black metal mask with 96 6mm holes arranged in diagonal lines. Each diagonal line is backed by a theatre quality red or blue or green acetate light filter. Behind the RGB filters are 96 light dependent resistors (LDRs) providing analogue output to 6 Arduino Megas. Each Mega supports 16 analogue inputs, so there are two for each colour.

The (7th) command Mega, collates the input from the 6 others, assigns percentage values for each RGB colour component, and compares these to known values. It also controls speech synthesis from stored allophone values.

The voice is created by an SP0256 “Narrator”. An obsolete chip by today’s standards, which assembles recognisable words from allophone sound fragments. I love this “War Games” type voice, and I use in several of my robots.

I included the 3 now-empty DIL sockets on the interface card for latches, but it turned out I didn’t need them.

As I mentioned. Fairly useless, but hopefully engaging nonetheless…

My thanks to

Tom Halligan

for inspiring me to post this!