Date: Wed, 4 Jan 2006 22:35:22 -0600 (CST)
Subject: mockup design study
X-UID: 118
Content-Type: IMAGE/JPEG; NAME="img2517.jpg"
Content-Type: IMAGE/JPEG; NAME="img2518.jpg"
Content-Type: IMAGE/JPEG; NAME="img2522.jpg"

All devices are set on the vehicle platform except for the headlight. I 
forgot about it and don't want to go back and re-shoot all the photos. 
Anyway, this gives a better idea what the robot will look like. Here is a 
manifest of the objects in the photos.

- vehicle platform itself (including steering motor, drill motors,
   drivetrain and mounting rails for electronics and battery packs)
- electronics enclosure (computers and motor control)
- two 18 volt 3.5 Ah NiMH packs
- 24 2300 mAh AA batteries in three 8 cell packs
     - one pack for electronics cooling fan and steering motor
     - one pack for the Netgear WiFi router
     - one pack for the computers and motor control board
- laser line level
     - two Sharp IR distance measuring sensors mounted on the front with
       crossing coverage fields
- Futaba GY240 piezo gyro
- Garmin GPS 18 LVC
- Maxstream XStream 900 MHz 9600 baud radio (in Penguin Peppermints tin)
     - dipole antenna strapped to top of plastic mast
- two ov51x based webcams, one facing forwards and one backwards

The mockup helps working out spatial and aesthetic relationship between 
different elements that were never designed to fit together. Until setting 
everything (except for the headlight) together, I couldn't quite see how 
it would all fit.

When I took architecture in high school, we had to make models of the 
buildings we designed. A design changed when it moved from two dimensional 
floorplan into three dimensional foamcore and cardboard. What appeared 
workable on a drafting table often became ridiculous once represented as a 
model. Mr. Aononsen's eye must have immediately recognized what was wrong. 
But most students could not perceive this until after fabrication of a 
model.

I'm not sure if even CAD visualization quite captures the nuances. My 
experience with CAD early in this project is not creative or time 
efficient. It is more of a documentation tool after the spatial 
relationships are already well understood.

Now that DARPA has all but announced an urban robot competition, the focus 
of this robot is as a technology testbed for this. My experience driving 
and cycling on high speed roads is that view to the rear is necessary. 
This robot will have a maximum speed of around 20 mph, roughly the same as 
a bicycle with an average human cyclist. So it needs to see cars that are 
overtaking.

No, I'm not so crazy or irresponsible as to do anything nuts. But I'm 
working through the design space with this robot. The primary objective of 
this project has always been education and preparation for the next 
generation that follows.

In truth, the onboard computers have insufficient performance to perform 
stereo correlation in real time. Monocular cues such as optic flow and 
feature detection might work. So there isn't much benefit to facing both 
cameras in the same direction.