Integrating Open Source Technologies: Robot control with LinuxCNC, AI, and EtherCAT Integration
Roque Barrios, Jaime Javier (2023)
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2023121537721
https://urn.fi/URN:NBN:fi:amk-2023121537721
Tiivistelmä
This thesis is dedicated to exploring novel approaches in utilizing the Festo iCIM 3000 robot within the LinuxCNC framework and incorporating Artificial Intelligence. The aim was to establish a basic framework for a Robotic Control System using open source software, with the objective of developing a comprehensive solution that addresses concurrent hardware and software challenges. The work encompasses detailed exploration and implementation of EtherCAT technology, open source tools, and the integration of voice recognition to move the robot that can be easily extrapolated to others as well.
The research focused on developing an advanced robotic control system using open source software. It involved single board computers for hosting, System on Chip as an EdgeAI solution, Linux kernel patches for real-time capabilities, and ROS for fluid communication with the robot. Evaluation of different EtherCAT masters was conducted to determine the most suitable for the project.
Custom implementations included integrating ROS into LinuxCNC, enabling TwinSAFE, modifying Whisper.cpp for system commands, and creating EtherCAT terminal drivers. Post-Training Quantization reduced the size of the Whisper model, enabling faster iterations. The study successfully achieved integration and optimization, showcasing enhanced flexibility in the LinuxCNC and iCIM robot system.
The proof of concept demonstrated full control of the EtherCAT terminals, enabling automated relocation of items, hand gesture recognition, gamepad control, and interfacing with external controllers. The successful integration and optimization represent a significant milestone in advancing robotic control systems, highlighting the flexibility and adaptability of the integrated LinuxCNC and iCIM robot system.
Overall, this work contributes to the fields of industrial automation, robotics, and AI by providing a comprehensive integration framework that can be used in production, vocational schools and universities.
The research focused on developing an advanced robotic control system using open source software. It involved single board computers for hosting, System on Chip as an EdgeAI solution, Linux kernel patches for real-time capabilities, and ROS for fluid communication with the robot. Evaluation of different EtherCAT masters was conducted to determine the most suitable for the project.
Custom implementations included integrating ROS into LinuxCNC, enabling TwinSAFE, modifying Whisper.cpp for system commands, and creating EtherCAT terminal drivers. Post-Training Quantization reduced the size of the Whisper model, enabling faster iterations. The study successfully achieved integration and optimization, showcasing enhanced flexibility in the LinuxCNC and iCIM robot system.
The proof of concept demonstrated full control of the EtherCAT terminals, enabling automated relocation of items, hand gesture recognition, gamepad control, and interfacing with external controllers. The successful integration and optimization represent a significant milestone in advancing robotic control systems, highlighting the flexibility and adaptability of the integrated LinuxCNC and iCIM robot system.
Overall, this work contributes to the fields of industrial automation, robotics, and AI by providing a comprehensive integration framework that can be used in production, vocational schools and universities.