For 2003, there are a lot of changes.  The camera from 2002 has been replaced by a CMUcam which, again, is just used to find the candle.  The CMUcam is mounted behind the aluminum sheet which covers all but the lens.  The aluminum also allows mounting of a D12 sensor on the top and an SRF04 sonar on the bottom.  All three are on top of a servo which allowed panning 90 degrees left and right.

At the left and right sides, there are D12 range sensors pointed forward and rear.   These were used just to detect walls in front of or behind the robot.  This was the new expert class with unknown configuration and the rear mounted sensors allowed the rear walls to be mapped without having to turn around to look at them.

The fan has been moved to the rear so its propeller would not interfere with any of the forward looking sensors

The rear view shows the rear of the CMUcam better.  The fan is mounted in a cutout of the front platform so that it does not go outside the 10 inch robot diameter when rotating.  The small circuit board to the right of the fan is the driver circuit described elsewhere.

The wireless breadboard in the center is attached by flat cable to the I/O pins of the processor on the platform below.  This allows connection of almost all the top platform components ON the top platform.

You can see the cable bundle going to the scanning sensors on the servo.  This is the bundle that was not properly clamped at each end and flexed the camera power wire until it broke.  You can also see the one wire outside the bundle that is the wire to the D12 sensor that I had to replace during the competition.

Note that I had added shrouds to most of the D12 IR range sensors to minimize the interference of their light emissions on the scanning D12.

I mounted most of the D12's vertically this year.  Vertical gives a little better signal accuracy...particularly when scanning across a corner.

The Left side has little new info.   The black project box just above and to the rear of the wheel is the starting tone detector circuit.  My new Adapt9S12 processor board is mounted on the solderless breadboard on the lower platform.

Right side

The bottom.  This photo shows my original use of omni-wheels for casters; until I found out they had a lot of drag moving sideways on carpets.  They were later replaced with skids made out of teflon.

I used Pittman motors driving wheels on each side through RC car gears.  This setup is described elsewhere.  Since there was no room for a spring assembly under the rear caster (as in previous years), I went to a torsion spring setup.  The rear caster is hinged just behind the batteries (in the center) and the wires going sideways from the hinge to the side wheel/motor mounts provide the torsion.

It was difficult to fit these large motors and wheels into a 10 inch diameter (I try not to be any bigger than this), but it worked out pretty well with room between the motors for the batteries. 

And after three years, I finally figured out that the easy way to hold the batteries in place is to use velcro strips.