Version 9, last updated by dbenamy at April 27, 2010 05:54 UTC

2009 Competition

We improved a number of things but didn't do very well.

Improvements:
  • 2 cameras- wider field of view.
  • Wheels powered separately. Fixed 1 wheel spinning problem.
  • (Attempted) better speed control- Proportional Integral Derivative control system for drive power. Increased shaft encoder resolution- not good- stamp doesn't sample fast enough at high speeds.
  • Pid2- a second interface board (custom pcb) using an Atmel AVR with code written in C. Lowered steering pwm period- stronger steering.
     
  • Moved shaft encoders to pid2 to increase polling frequency, but it looks like it still can't poll fast enough for good readings at high speeds, even with a low resolution shaft encoder.
  • Improved cone avoidance.
  • Fixed gps.
  • Fixed compass.
  • Gui got map (and other changes)- graphical representation of path, heading, and way points.

    (cameras are messed up in this picture)
  • Optimized graphic processing code which improved frame rate of vision.
  • Switched to mercurial for revision control- allowed normal working in the field- very useful.

The problem was that the color calibration was off so the robot thought some bushes at the beginning were mostly road. It got too close too them and the sonar sensors detected them as an obstacle and stopped the robot. For fun, we moved it past that point and it did the rest of part 1 correctly?

For next year, we're considering creating a new robot base which is smaller. This will make transportation and steering easier. It will probably also break everything :-) We're also kicking around the idea of rewriting the code, possibly using ROS. We really need to solve the color calibration problem and it would be great if we could work out a better way to test everything without being at the course.

2008 Competition

We competed again at the 2008 Mini Grand Challenge on April 26. We did better than last year, but so did the competition and we took 3rd place once again. We managed to get in 2 runs this year. The color calibration somehow got screwed up for the first one so the robot thought everything was road and drove off the side of the path a bit after the first way point. It was actually at the bottom of the downhill where we stopped last year so even our bad run this year was better than our (only) run last year. On our second run we fixed the brightness and our robot made it most of the way through part 1. We got going too fast on a downhill just before way point 4 and drove a bit off the side of the road and got stuck. The anti-stuck code actually kicked in, but it wasn't quite enough to get us back onto the road.

Unlike last year, our hardware was very stable and didn't give us any trouble. This was probably because we brought spare parts for most things. We're getting this robot base into good shape. If we stick with it for next year, we want to increase the angle of view by adding another camera or getting a fish eye lens for this camera or a different one. We'll also need to get the compass working. And we'll probably want to power the wheels separately. With them in series, one wheel can spin while the other is stuck.

We gave up on the Jack Bauer theme. Instead we found a ridiculous mannequin, named it Jack, and stuck it in the robot. It also spoke using text to speak (MS Mary, I think) instead of the 24 sound clips.

Changes from last year include: working GPS, not quite working digital compass, better sound system, shaft encoders, variable power motor controllers, code to do speed control, orange cone growing, text to speech, better stopping when something is in the way, better camera mount.

2007 Competition

We entered the Jackbot in the Mini Grand Challenge competition held at Penn State Abington on March 31, 2007. Despite a bunch of problems we encountered on the day of the competition, the Jackbot finished in third place. The Jackbot was named after Jack Bauer from 24 (and yelled Jack Bauer quotes) in the hope that it could handle any situation. That didn't work out too well.