/VAST

VAST - Valid Action State Transition Optimization for Autonomous Robots

Primary LanguagePython

VAST - Valid Action State Transition Optimization

The idea of VAST is to provide all the resources you need to ML and RL running on your robotic projects. Each master folder contains a robot prototype. These folder contain setup intrucstions, wiring details, code examples, results and more.

Example of VAST Controller

Nerual Network Visualization

Visualization tool I create to help visualize how the weights and biases of NN change during training.

Click here to see example code

Example

Snakebot - Master

  • add folder with all Qlearning and DQN files and documentation once verified and completed, highlight plots, images and video results here
Click here to see example code

Snakebot Prototype

Fishbot - Master

  • add folder with all Qlearning and DQN files and documentation once verified and completed, highlight plots, images and video results here
Click here to see example code

Fishbot Prototype

Intellegent Locomotion Controller - Master

  • add folder containing all files and documentation for developing an Intellegent Locomotion Controller which uses
  • Qlearning to devleop low level control polcies which are then saved and used by a highlevel DQN network as actions
  • in order to naviage to difference global coordinates highlight plots, images and video results here

VAST App - Master

  • add kivy documetation and files for creating a monioritng and deployment app

ROS Developer Guide

  • code details and setup for using ROS and online development studio