Work of Comp 396 Undergraduate Research Project
The 2.5D simulator works in conjunction with the general purpose AI. It has two purposes:
(a) it is a tool used by scientists who want to build a simulated world, with objects and NPCs and robots.
(b) it permits a unique general purpose AI to be assigned to each robot.
The physics of the robots and the world need to be acurate enough so that what an AI might learn in that environment can be transferred to the real world.
The graphics of this simulator is not important. The physics is important.
The ability for a non-programming scientist to use it as a tool where they can setup an experiment, watch it play out, and save results for futher analysis.
The simulator should also have a "play back" feature to explore events.
The tool should also have a "look into the brain" feature to explore what an AI was thinking at selected events.
2.5-D Simulator Project
Task distributions:
This project is divided into 3 clear main parts, with the world physics of the simulation (including interactivity between the avatar and the environment),
Graphical User Interface and the Artificial Intelligence API. Below, we will summarize what area each person will work on individually.
At the end of the development stage, we will work on the optimization of the code and program execution together.
Input: There will be a predetermined configuration file deciding the physics of the simulated world and the pathname to the AI database for the avatar and NPC. Inside of the GUI, the user would be able to set the exact configuration that they want to have in the simulation such as the terrain, shape and objects in a room, number of NPC’s, starting location of the avatar, etc.
Output: Each simulation will output a CSV file for further analysis.