Overview / Usage

The OmniSTARK project aims to create a perfect platform to investigate self-driving intelligent software agents. The robotic platform conceptions and design are personal implementations and intend to leverage the mecanum wheels design to achieve seamless omni-directional movement to navigate and explore complicated environments.

Methodology / Approach

The project is initiated from scratch, beginning all the way from conceptualizing and assembling the components all the way to code development and implementations of navigation techniques with RL agents.
We start by test driving the omni-directional movement of the platform and basic controls required to achieve it on the BLDC motors using Reinforcement Learning techniques as a controller.
A vector of sensory inputs (Ultra-sonar or laser) consists the starting point of the environment mapping and decision making for navigation purposes.
The platform would present a solid development space for further applications on SLAM AI and DL techniques to reach high flexibility movement in densely crowded environments.

Technologies Used

The project is based on the following technologies:

• Mecanum wheels
• Reinforcement learning Agents implemented in Intel Python
• Intel Coach Library
• Up squared Dev Board
  Potential future technologies include:
• Intel RealSense camera module
• Intel Movidius NCS

Other links

• OmniSTARK project website