- Implement NVIDA AI packages with e.DO Robot in ROS2
- Use live classification to detect different colored cubes and buckets and sort them. Also use gesture input to control the e.DO with different hand gestures.
Supervisor: Kamel Rushaidat, Ph.D. https://www.linkedin.com/in/kamelr/
Allen Paul - Team Lead
Khalid Awad - Implementation Lead
Laura Gipson - Documentation Lead
Sunny Prajapati - QA Lead
Wameed Miriako - Integration Lead
There are tutorials to help set up this project located in the tutorials and documentation folder. The following videos also provide examples of how to build and run this package.
Demo and Explanation of MoveIt! Integration: https://youtu.be/PFj-qI9juJk
Demo of sorting: https://youtu.be/KuTI88DBrp0
Explanation of sorting: https://youtu.be/jmedsHACDJY
- ROS1 Melodic
- ROS2 Eloquent
- Installation of the ros1_bridge package
- edo_core_msgs package, along with the updated, ROS2 friendly, version of the package included in this repository
- ncurses
- Pytorch
- torchvision
- timm
- ros2_torch_trt: https://github.com/NVIDIA-AI-IOT/ros2_torch_trt
- vision_msgs: https://github.com/ros-perception/vision_msgs/tree/foxy
- Convert wrapper class for e.DO robot previously developed by Jack Shelata Link to Jack's GitHub
- Confirm manual control capabilities on the COMAU e.DO robot using the NVIDIA Jetson control module
Ayush Shrestha - Team Lead
Scott Howard - Implementation Lead
Adam Erdman - Integration Lead
Seth Buchinger - Documentation Lead
There are tutorials to help set up this project located in the tutorials and documentation folder. The following videos also provide examples of how to build and run this package.
Building the package: https://youtu.be/r9rDI4yP-Us
Using the package: https://youtu.be/egTKGkaJMBs
Explanation of code base: https://youtu.be/z64mXXk3P70