Short course · 3 or 5 days · morning theory, afternoon labs on SO-101 arms
Robotics and AI: industry intensive
A hands-on 3- or 5-day robotics and AI course for engineers: transforms, kinematics and control, camera perception, imitation learning, ACT and diffusion policies, VLAs and edge deployment on SO-101 arms.
Who it is for
Engineers who know Python and want to build robot-learning systems: from teleoperation and data collection to trained policies running on edge hardware.
What you get
- One leader and follower SO-101 arm pair per two to three participants
- Teleoperate, record a dataset, train and evaluate policies in one week
- ACT and diffusion policies compared; a vision-language-action model fine-tuned
- Edge deployment with latency and safety measured
- Optional extra day: FPGA track for hardware teams
Before the course
A working Python 3.10+ installation on a laptop. The 3-day version uses pre-assembled, pre-calibrated arms, a pre-recorded fallback dataset and pretrained checkpoints.
Daily schedule
| Day | Morning: theory | Afternoon: lab |
|---|---|---|
| 1 | Robots, AI, transforms and forward kinematics | Assemble, calibrate and teleoperate the arms |
| 2 | Inverse kinematics, control and cameras | IK reaching; camera calibration |
| 3 | Perception and imitation learning | Record a 50-episode dataset; ACT trains overnight |
| 4 | ACT, diffusion policies, RL and VLAs | Evaluate ACT; fine-tune SmolVLA |
| 5 | Deployment, FPGA and safety | Edge deployment; team demos |
Lecture 0: Your course guide
The 19-slide student guide: how the course works, what you will learn, the 12 chapters, labs, hardware and the team project, how you are assessed, and how to get ready for Week 1.
PDF · 19 slides
Interested in running this course for your team? Get in touch.