The melexis safety trophy is an international competition where perticipents are required to build an autonomus robot that will navigate an unknown course avoiding hot cans and knowcking over cold cans while not crossing the lines or bumping into bricks, it completes the course my locating a flashing beacon.
The robot can't be remote controled it has to rely sensors to detect its surrounding, from the beginning the aim was to keep the robot simple use the bare minimum of sensors to complete the task. It needed proximity sensors to detect the obsticals, line detectors to locate the lines on the road, a beacon detector to detect the flashing beacon and hopefully a temperature sensor to detect the hot cans.

The robot was built from layers made out of perspex on each layer were different sensors, batterys, motors or control circutiery.

The motors were going to be 2 modified servos but these proved to be unsuccessfull, we then decided on larger cordless drill motors (these came from the robot wars vehicle), these were attched to the bottom layer. These are high power motors drawing about 10A at 6v, 2 6v gell betterys were chosen to power them, these were wired in parralel. The first attempt at constructing a motor controler was unsuccessfull; 2 h-bridge ic's were chosen (looking back now they were very underpowered). I wasnt able to made the circuit work so that idea was scrapped, I used the same style of controler as on the robot wars vehicle which is allways reliable, switches; this time high power relays rather than manule swithes controled by a servo. These relays were then switched by another 4 relays which were controled by 4 transistors; I have no idea why I used such a complicated system.

The motor transistors were then controled by a 16f877 PIC which took inputs directly from the proximity sensors beacon detector and line detectors.
When testing the beacon detector I couldn't reliably detect the beacon from more than a few metres away, this ment that it wouldnt know which way to go, to solve this problem I installed a compass a CMPS03 (I think thats the right number, I will check sometime), this then feed its PWM output directly into a second PIC which decided if the robot had to turn left or right.

During testing (a week before the competition) the line detectors were found not to work (bad deisgn), so they were programmed out, we would just have to hope it didn't cross the line. The beacon detector then wouldn't work properly; in hindsight it was a good deisgn and would have worked if I had spent more time on the programming.
This ment that the robot was down to proximity sensors and a compass.
A basic program was written that ment that the robot went roughly in one direction but avoiding obsticles, this worked reasonably well except that in some places the cold cans (we were supost to knock these down) were placed closely together, this meant that the robot couldn't get through.
I then changes the program so that if it detected a single object, i.e. one sensor, the robot would carry on but if it detected something on 2 or more sensors (a brick) then it would avoid it. This worked verry well and the robot completed the course in a decen't time.

Problems later arrose when the course got harder and cans were packed closer togther meaning that it couldnt distinguish between a brick and a can, it eventualy got through after spending some time looking arround.

The competition went well and the robot performed well, it completed all the competitions in good time, after each couple of rounds robots were cut but we stayed in untill everything went wrong.
During the last 3 or 4 rounds we had problems with the batterys, it was draying so much power that the batterys were being drained quicker than we could charge them (and no we didn't have a spare set), I eventualy resorted to dismantleing the battery from the cordless drill I took, and for the last round the other Brittish team lent us a bettery which we hooked up.
Unfortunitly in the 2nd to last round and one of the motors started playing up, it had almost no power which ment it stated to go arround in circles. after this round it was tested and found to work so we put it into the final, it was just down to me and anoter guy with a logo robot who wone last year, it was very exciting.
You guessed it we didn't win, the robot traveled about 5 metres before the motor stoped working and the battery died, when they added up the scores it was very close, the other guy got further but he didn't complete the course and he had to interveen with his robot so he had time added.

I didn't win but came second and got €3000 worth of vouchers with a european electronics company, which I unfortunitly had to share with the person I was working with even though I built and programmed the entire thing my self, i'm not bitter!