Since last year, the robotics industry has been booming, with significant advancements in AI and machine learning. New concepts and prototypes are emerging rapidly, making it an exciting time to explore this field.
The life for a single person can be quite busy, it’s really frustrating to come home after a long day and still have to do chores like cleaning, cooking, and laundry.
My current job is flexible, but lacking innovation. I want to work on something that challenges me and allows me to learn new skills. I am approaching my 30s, and I don’t want to end up a day in my 80s, with the same working routine, which is happening to many people in my family. I know that we are undergoing another industrial revolution, and I want to be part of it.
Well then, why not give it a try?
LeRobot - aims to provide models, datasets, and tools for real-world robotics in PyTorch. The goal is to lower the barrier to entry to robotics so that everyone can contribute and benefit from sharing datasets and pretrained models.
XLeRobot: Accidentally I came across this project that’s called XLeRobot and was at once fascinated by it. It combines several existing LeRobot projects, such as the SO101-Robot arm and the LeKiWi moving base, with an IKEA RÅSKOG Trolley to create a mobile robot platform that can navigate and manipulate objects in a domestic environment. I sent immediately a direct message to the author of the project, @VectorWang. He answered me very kindly and we found out even some mutual friends. I didn’t hesitate, but ordered the IKEA trolley and started to collect all the necessary parts.
The LeKiWi trying to catch a Kiwi.
As the first step, I would follow up the tutorials and replicate the robot in my home environment.
A very obvious problem in the current design is that the moving base, which is driven by 3 omni-wheels and designed for a single robot arm, provides too little momentum and ground clearance, which constrains the robot’s ability to move around in a typical home environment. Therefore, I would like to replace the moving base with a more robust one, such as a tracked base or a four-wheel drive base.
A possible solution would be this Chasis from Taobao:
Here is a timeline I have in mind:
Skills and knowledge I would need to acquire: - Robotics: Kinematics, Dynamics, Control Systems - Computer Vision: Object Detection, Image Processing - Machine Learning: VLA Models, Reinforcement Learning - Hardware: RGB-D Camera, Raspberry Pi, Motor Controllers, Optional: Jetson Nano - Software: ROS, Python, PyTorch, OpenCV
The first procurement has been done on 2024-06-25 according to the Bill of Materials.