As a semester project for my master’s degree in Robotics, Systems and Control in 2019, I had the opportunity to work with Mobmi, a mobile manipulator robot of the Autonomous Systems Lab (ASL) at ETH Zurich. Mobmi is composed of two main components, a Clearpath Ridgeback Mobile Base Platform and an ABB Yumi. Ridgeback allows Mobmi to move freely in 2D space thanks to its omni-wheels. Yumi, which is mounted on-top of Ridgeback, has two arms with 7DoF each and a range of roughly 0.6m. The goal of my project was to adapt the software running on Mobmi to let it combine multiple tasks (like driving to a specific location, grasping objects etc.) in a simple manner.
The output has then been presented by combining all tasks necessary in order to serve a coffee fully autonomously.
This project allowed me to improve my skills in ROS. Especially, I was able to get familiar with 2D path planning concepts for finding collision-free paths in unknown environments as well as pre-recorded maps and 3D trajectory planning for grasping objects with orientation constraints using the MoveIt! motion planning framework. I will be using the newly learned knowledge in my autonomous R2D2 project.
I recorded the result of my work in a One Shot together with visualizations of what the robot sees: