Unlike last year where the robot had to avoid cans and get to one end of the course this year the course is a lot more complicated, the robot has to to identify road signs, stay within the white lines and avoid other cars; the hardest of these tasks is to identify the road signs.

Traffic Signs
This is the hardest task by far, there are 4 different road signs and the robot has to recognise each.
I first considered using a camera and image recognition, I found a simple java application that would identify a road sign based upon previous stored images, this would require some serious modification of the program, it would need to access the computers parallel or serial port and output information as to what type of sign it was. The only experience i have with Java is an introductory course i took in my first year, with this in mind I quickly ruled out this idea.
The other idea I had was based upon a laser scanning system used by the other British team last year. This system involves a laser which scans across the road sign and a predefined height, as it scans across a detector looks as the intensity of the reflected signal, the detector should be able to pick up the change from blue to white in the sign and determine what it is. It order to eliminate ambient light the laser would output a modulated signal the detector would then only look for this modulated signal, it would then amplify the signal and integrate it this act as as basic edge detection. The detector consists of a single pixel red light detector, this sits at the end of a tube with a focusing lens mounted at the front, this is calibrated so that at the correct distance the reflected dot is on top of the detector. the detector is then connected to a filter to remove any low frequency signals i.e. mains lighting and then to an amplifier to amplify the very small reflected signal.
The detector would run autonomously on its own PIC, that is it wouldn't wait for the master controller to tell it to look for a sign, a proximity detector would be mounted on the laser this would detect the presence of a traffic sign and start the detection routine.
The road signs that needed to be identified are blue with a white arrow in the middle, if the correct position on the sign is chosen the horizontal change in colour from blue to white is different for each of the 3 different signs, the laser would use the change in intensity from white to blue to identify the sign.

10/02/05
I have built the sensor module and I have been mounted it to the laser; I have tested it with the laser connected to a square wave generator and the detector to a high pass filter and the initial tests are very good, the changing signal is very small but this can be simplified, I am still trying to find the time to test a more advanced filter and amplifier setup.

Proximity Sensors
The robot needs to be able to detect any other cars on the road as well as any other obstacles, to do this i am mounting various proximity sensors around the edge of the car, I have decided on the Sharp GDP15? digital proximity sensor. This device is pre calibrated to a distance of about 26mm?, when an object gets closer than 26mm the output goes low. I will have 3 of these on the front of the robot and others around the sides to detect the other car as the robot passes it. These sensors will all feed into an 8 bit port on the sensor microcontroller.

Line Detection
Detecting the white lines is an important job, this will keep the robot on the road and going in the correct direction, unlike last year when i made line detectors out of comparators LDR's and LED's I have decided on a complete device.
I will mount several of these down the sides, these output an infrared signal and look for the reflection, the black road should absorb the signal while the white lines should reflect it.