Inspiration
Making a valuable impact on a real world application using AI with robotics.
What it does
Effectively utilizing a VisonLanguageAction Model concept to orchestrate atomic actions of 6 DOF (degree of freedom) robotic arm to demonstrate pack and assembly task done by humans in a warehouse.
How we built it
- Used Phospho AI recording tools, capturing the Arm actions and create a library of actions to be orchestrated by VLA, Used Phospho APIs to move the arm orchestrated with an LLM to perform the tasks. The input is a QR code that can be read by the robotic arm to initate the assembly task, we used OpenCV for this task
Challenges we ran into
- Creating the staging environment to capture the required workflow with limited reach of the arm
- Lack of time to train.
Accomplishments that we're proud of
- Fully encapsulated project and working demo of a real world application
What we learned
- Using Teleoperated actions to capture datasets that can be used by a VLA.
What's next for Atomic Kitting
- Explore adding more autonomous capabilities to improve the dexterity of the arm for grasping objects
- Include exception handeling capabilities
Log in or sign up for Devpost to join the conversation.