Progress
As of 5/6
The project is done. As it turned out, we were overly optimistic in
attempting to get multiple robots doing cooperative mapping. We got a
single robot to map out its environment, using FASTSLAM. The
following webpages detail the end results of the project: Victoria's landmark detection code, Paul's FASTSLAM code, Steph's wandering and control code.
Videos of the robot in action are here and here.
Victoria has also completed the picobot rules, solving map 0, map 1,map 2. Victoria left the solutions to the path planning
problem outside Prof. Dodd's door Friday evening.
As of 4/21
Steph has updated the wandering client
to use the new mapping and vision code, as well as making several other
adjustments.
As of 4/13
Paul has changed our maptool code so that it understands landmarks. Victoria has one
copy of the vision tool which identifies
artificial landmarks and estimates their size, and one copy of the
vision tool which plays Set!
As of 4/8
We have a shiny new
master plan for the semester of what we want to get done.
As of 3/25
The MCL code is now working fairly well.
In addition, we have a plan for the next
two weeks, focusing on tidying up loose ends from the first part of
the semester, and preparing for coordinated mapping using wireless.
As of 3/10
We have an implementation of Monte Carlo
Localization (MCL) mostly working. However, Paul is out of town at
a conference, so it's not clear if we will be able to finish in time.
Week of 3/6
This week was spent getting wandering working properly. Steph
developed a guided client for moving the robot under manual control (code here), and a wandering client which follows the left
hand wall.
Week of 2/27
To start off this week's work, Victoria mounted the sonar unit Prof. Dodds
created for us on our robot. Then, Paul and Steph started writing a client to control the robot
so it could wander the halls. Unfortunately, Strider, Steph's
laptop, refuses to talk to both the sonar unit and the robot at the
same time, so our wandering behavior has not yet been tested.
Week of 2/20
For this week, we've completed two separate tasks: editing the image processing software Prof. Dodds
gave us to detect red moldings , and writing a remote control
client for our robot, so we can more easily control it while it roams
the hallways.
Week of 2/7
This week, we completed two separate tasks: soldering and connecting up a servo motor, and testing the accuracy of the robot's odometry.
Week of 1/31
At present, we have two robots, unimaginatively named One and Two.
Paul built Robot One and Steph built Robot Two. We tried
to make the robots go places, but instead, we got an error!
