Advanced ML Agents

Manisha Biswas

Manisha Biswas

Kolkata, WB

We would be creating smarter game agent simulations for Unity Game IDE embedded with Open vino toolkit ...learn more

Project status: Under Development

Game Development, Artificial Intelligence

Intel Technologies
OpenVINO, Intel Opt ML/DL Framework, Intel Python

Links [4]

Overview / Usage

What we are going to do
i)Improving Reinforcement Learning training and inference
ii)Creating Simulation with advance graphics in the PC form factor
iiii)A highly simulated environment for rapid simulations
iv)Imitation Learning basics

Methodology / Approach

How we are going to achieve it
As we are using Game engine unity we are bound to get good graphics for the CPU

For agents that we create for the unity we will make it more faster as the final results for the simulation using open Vino will be reflected in the scene.

We would be using unity ML Agents V0.4 for our training too we would be using the open vino toolkit as well as Intel Optimized Python for generating the trained file that will be consumed as tf model.

That TF model we will optimize using the optimization methods of Open vino toolkit and will showcase the entire simulation in a screen that will be rendered from PC.
Advantages of Using Unity ML Agents
Unity ML agents environment has a full python support so for advance implementation of the inference model we can use the python API from Open Vino toolkit for more rapid simulations.

Example environments for better simulation are there in the open source code we can use those lot environments.
Using Unity ML Agents we get access to multiple environments and training scenarios when applying open vino toolkit we can definitely tune in the hyper parameters for better simulations and varied results.
for Self driving simulation Unity ML Agents have built in support for imitation learning .Applying Open Vino Toolkit helps in producing the better results.
For Optimization we might use the Following method for creating customized layer for the tensorflow model:
For a tensorflow Model
First we will have to initiate the setupvars.sh script.
We will build inference engine layer with tensorflow runtime.
./tf_call_ie_layer/build.sh

Technologies Used

Intel optimized python
Open Vino toolkit for inference
A PC powered by Intel Architecture

Collaborators

There are no people to show.

Comments (0)