CDS 101/110a, Fall 2006 - Course Project
For the project track of CDS110, each group will be working with two controllers.
First, you will be choosing appropriate gains for a pre-existing controller called follow. The idea of this is for everybody to learn how to run code on Alice and collect data. See here for more information on follow, and here for the steps you need to complete before you will be allowed to try your gains out on Alice.
The second half of the project will be designing your own controllers, based on the project you chose. For these, you can assume that the sensor data will be made available to you -- However, you will need to think about what data/format you need. More information about this will be posted later, and you should feel free to ask Laura for more details.
note: I'd prefer to receive questions by email or during the recitation sections. I will answer any emails w/in 24 hours (usually much faster, but sometimes I am swamped with other classes) -Laura
Field Test Signup
- If you need to test this week, emailed me your schedule, and none of these times work for you, let me know (sooner rather than later) and I can try to work something out -Laura
Thursday, November 16
Friday, November 17
- Recitation Section: How follow works (pdf)
- What needs to be done for this week: GOTChA charts
- First cut simulation
- Recitation Section: Prep for Alice demo
- At some point this week (times TBA) each group will need to schedule time in Alice to learn how she works and to test out their gains in follow.
- Start working on first cut simulation for project; if you've tested controller in simulation, you will be allowed to test on Alice @ the demo.
- Recitation Section: Presentations of each group's state-space implementation
- It is possible that the first attempt at designing a stable controller with acceptable performance won't work. There may be an oppportunity for a second test.
- Each group should prepare a short presentation describing the controller they designed. This should include charts of: controller design, pole placement, data, etc.
- Recitation Section: How trajFollower worked (the controller that we used for the 2005 race)
- At this point, everybody should be working exclusively on the separate projects. Questions that you should be thinking about include:
- What inputs do we need? What format? How can we get them? (Talk to Laura - she will help with interfaces/spoofed data)
- Preliminary control law design?
- How can we test our controller in simulation/real life w/out putting Alice in danger?
- Recitation Section: Discussion about everybody's current status
- Recitation Section: Group presentations
- Lateral control in forward and reverse, with gain scheduling (DGC) - Design a controller that matains the lateral position of the vehicle when it is moving either forwards or backwards and schedules the gains based on speed. This requires a rederivation of the dynamics of the vehicle that model the motion when moving backwards and determining how to switch between forward and reverse motion. The specifications for the controller will be based on the needs of the 2007 Urban Challenge.
- Kenny Oslund (75)
- Christine Parry
- Mahipal Raythattha
- Andrew Krause
- Craig Montuori
- Dominic Rizzo
- Lane following (DGC) - Design a controller that uses information about the right and left lane boundaries to remain centered in the middle of the lane. The location of the left and right lane boundaries will eventually be given by a vision system, but for the course project simulated lane boundaries will be used.
- Chris Schantz (75)
- Johnny Zhang (75)
- Steve Gray (75)
- Mohamed Ali
- Gustavo Costa
- Tom Duong (75 only)
- Stopping at a line (DGC) - Design a controller that brings the vehicle to a stop within 1 meter of a stop line painted on the ground. The location of the stop line will be a combination of GPS data (when far from the stop line) and vision-based data (when near the stop line).
- Jessica Gonzalez (75)
- Francisco Zabala (75)
- Russell Newman
- Sanjeeb Bose
- MIke Ikeda
- Harrison Stein
- Reactive obstacle avoidance (DGC) - Design a controller that uses direct information from a laser range finder (LADAR) to bring the vehicle to a stop and/or swerve to avoid a "pop-up" obstacle. For CS/EE/ME 75 students, this could be a good project for someone on the sensing team.
- Ghyrn Loveness
- Klimka Szwaykowska
- Chris Yu
- Looking down a road (DGC) - Design a pointing system that can point a camera down a road to the left or right of a vehicle. The location of the road must be determined based on visual data (could be combined with a EE/CNS 148 project). For CS/EE/ME 75 students, this could be a good project for someone on the sensing team.