The project aims at testing machine learning algorithms for visual learning.
1. Creating a simple 3D environment with some objects.
2. An agent can be driven within the environment with basic interaction (pushing and pulling objects for example)
3. keyboard control of the object.
4. An agent-centric camera takes 2D frames of the field of view of the agent and provided as a 2D matrix.
5. movement direction, speed, rotation direction, and object proximity are available in input arrays.
6. Every time step, computation is made using the input arrays and matrix. according to the code that I will add. Possibly adding movement to the agent through the algorithm at a later stage.
7. Frames for display and visualisation of machine learning data and export of generated data in CSV or similar.
All of this in a desktop application and open source where I can edit the learning algorithms.
The application is for research purposes. It is expected that I will need modifications over the long term every now and then.
Additional help needed for providing me with help on using the tools needed to develop code in the chosen programming language. If professional game engines are used, there should be the ability to work on a free version and to be able to expand code (in form of matrix manipulation, data storage and retrieval, mathematical operations) without limitations.
If interested, please take your time to think about the project. Please clarify:
-Programming language proposed.
-3D modelling method, whether using your own code or using a 3D engine.
-How I can use the application to add my own code and use it for research.
I will not award this project in less than a week time so please don't feel hurried to bid and make sure this project is suitable for you.