arlk / robot-trajectory-tracking

Path generation with a cleaner and more beautiful interface with Processing & Python

Robot trajectory tracking

User-defined path generation : COMPLETED

L1 Navigation: COMPLETED

Obstacle Avoidance: PENDING

Path Generation:

This code provides a cleaner and more beautiful interface to work with than its MATLAB equivalent. It is written in the Python and blended with Processing, a development environment. It can be used in laboratories to define user-generated trajectories for individual robots. Implementation for swarm robots is still a work in progress.

This is currently on available to Linux (Debian) users - I am working on implementing this on other platforms including Windows. Before executing this program, you will need to download the Python module for Processing, which is avaiable in both a 32-bit and 64-bit versions.

You need to set the following environment variables before you run the script:

export PATH=$PATH:/PATH/TO/REPOSITORY/robot-trajectory-tracking
export PATH=$PATH:/PATH/TO/PROCESSING/processing.py-xxxx-linuxvv

Write these changes directly to ~/.bashrc and restart the terminal or source bashrc.

You're all set! Once all your variables have been correctly defined, all you need to do is execute the run script.

./run.sh

run.sh is a bash script that executes the required files, so the user does not need to run each of the files seperately. But the user-defined variables will need to be changed in each python script.

udps.py can simulate data instead of your mocap software, which is extremely useful if you need to debug your code before implementing them with real robots.

pathgen.py: Much like it couterpart this code assumes that you are receiving the following motion capture data via UDP:

1. x coordinate location of the robot in meters.

1. y coordinate location of the robot in meters.
2. yaw or orientation of the robot in radians, using the standard convention.
3. robot id specific for each robot.
4. status: 1 if tracked or 0 if untracked.
5. frame of the mocap data being sent.

Of course, the communication protocol can be changed to allow more variable such as z, pitch,roll to be obtained as well.

Specifications of the motion capture environment - room sizes, robot dimensions and colors can be specified at the beginning of code. And other UI-related customizations such as grid spacing, font sizes, spline smoothness etc. can also be found in the code.

Functionality:

1. Real-time reproduction of the scene from the motion capture system:

2. Drawing a trajectory with the cursor with immediate Catmull-Rom spline interpolation.

3. Clear trajectories whenever you need to and start again.

Output: The trajectories generated are stored as a comma separated values in files, referenced by their robot id.

###L1 Navigation:

multiL1.py can control multiple robots and make them track their trajectories independently and simultaneously.

Videos: 1. Single robot following a user-defined path.

Future work involves using these files for obstacle avoidance using L1 navigation.