Gradient Descent Pathing Algorithm Demo

This program is a demonstration of the gradient descent algorithm for pathing around obstacles in a 3D environment. It uses VPython to display 3D graphics and NumPy for matrix calculations. Implementation done in Python 3.8.7, no other versions tested.

To run from directory

1. Install requirements with "pip install -r requirements.txt"
2. Run "python animation.py"

Program use

1. The program will set up the inital scene based on lists cyl_pts and ball_pts (Note: Bind the function print_pos to a click event to test cylinder positions)
2. The program is hard coded to generate 6 balls, but this can be adjusted by adding the proper number of color values to the colors list along side updates to ball_pts
3. After initial set up, click anywhere in the bounds defined by the white rectangle to have all balls path towards it
4. When a ball reaches the goal, it will stop and wait for the other balls to arrive
5. The program will loop until all balls reach the goal (Note: Program status will be shown in a caption below the scene)
6. The goal can be dynamically changed both while pathing and while stationary

Known issues

1. Balls may get stuck if the goal is placed too close to an obstacle that directly blocks its path
2. Balls are repelled by eachother, which can make all balls reaching the goal difficult (Note: Place a new goal between the balls until the status is no longer pathing)
3. Placing obstacles too close to eachother will result in local maxima that balls will not be able to path through

References:

Algorithm: Cost Minimization for Animated Geometric Models in Computer Graphics by David E. Breen

Documentation for Vpython: https://www.glowscript.org/docs/VPythonDocs/index.html

Documentation for NumPy: https://numpy.org/doc/