Date: Fri, 17 Mar 2006 20:50:03 -0600 (CST)
Subject: driving in a parking lot
X-UID: 151

The camera is mounted on the robot where the 900 MHz antenna mast normally 
screws in place.

http://golem5.org/robot1/video/mvi2826.mpg (about 2 megabytes)

The three modifications for this run are:

1. steering bobbin has large aluminum washer to keep the cords from
    slipping off

2. heat shrink tubes hold the nylon cords on the Pitman arms

3. fiberglass mesh screen holds the cables away from the front wheels

It seems to drive o.k. I also learned more of the performance envelope. 
The robot is easy to roll. Early in the video, you can see the horizon 
tilt at about 25 degrees when the robot lifted off the left side wheels in 
a turn to the left.

The run was done entirely in low gear which is probably advisable until 
there is better control and knowledge of the safe performance envelope.

I thought it was strange that so many DARPA Grand Challenge vehicles 
rolled. But it is actually very easy to do. Careful driving is required to 
avoid a roll. The robot has a relatively tight turning radius.

The high pitched whine is the 2 KHz PWM to the motors. The rapid 
ratcheting sound is the slip clutch on the drill motors. The motors have 
to be started slowly or else wheel inertia causes the clutch to slip.

The next test drive is scheduled for Tuesday of next week. The primary 
objective is use of the 900 MHz radio instead of WiFi with proportional 
steering. That means a feedback loop on the robot that uses the 
potentiometer connected to the steering bobbin and the rate gyro to hold 
wheel position. The secondary objective is grabbing individual JPEGs from 
the network camera and converting them into bitmaps. This is necessary if 
the robot is to follow a road. The tertiary objective is mounting a laser 
on the robot for the nighttime run on Tuesday. This covers driving 
control, remote and onboard, as well as basic vision. I need to know how 
well the laser works at night as that will determine how to proceed with 
vision.

As I'm tired right now and have oncall production support duty from around 
1 am until perhaps 3 am, I'll be sleeping deep into Saturday. That leaves 
all of Sunday and then Monday evening to accomplish the objectives above.