Train and deploy whole-body control systems using Stanford’s Mobile ALOHA platform — a mobile manipulator robot designed for high-dexterity tasks in constrained environments. This course gives hands-on access to real mobile manipulation hardware, cutting-edge imitation learning techniques, and ROS2-based deployment pipelines.
You’ll go from dataset collection to autonomous execution — with code that runs on the real robot.
🧠 What You’ll Learn
✅ Whole-Body Control for Mobile Manipulators
Implement bimanual and base movement control strategies for real-world tasks (e.g., drawer opening, two-handed object transfer).
✅ Imitation Learning & Dataset Bootstrapping
Record demonstrations via whole-body teleoperation and train high-performing policies using supervised behavior cloning.
✅ ROS 2 Multi-Modal Perception Integration
Process 3D video, proprioception, and teleop motor control signals and train a transformer neural network for autonomous operation.
✅ Human-in-the-Loop Debugging & Autonomy Transfer
Refine learned behaviors through real-time evaluation and policy fine-tuning using open-source ALOHA tools.
✅ Final Challenge: Autonomously Execute a Real Task
Cook, clean, manipulate, or navigate: you'll deploy your trained model to execute a mobile manipulation task from scratch.
🧪 Platform Details
Hardware: Mobile ALOHA robot — bimanual arms + mobile base
Sim: NVIDIA Isaac + ALOHA teleop stack (custom tooling provided)
Sensors: RealSense RGB-D 3D cameras + joint encoders + mobile base odometry
Software: ROS2 (Foxy), Python 3.10+, PyTorch, Behavior Cloning codebase
GitHub Access: Full codebase and open dataset support
📅 Format & Details
Duration: 4 Days, In-Person
Location: Portland State University | Robotics Lab
Cohort Size: 10 Engineers Max
Robot Access: Hands-on with real Mobile ALOHA units (1 per 2 students)
Included:
12-month access to all code, videos, and tools
GitHub repository with reproducible projects
Certificate of Completion + ROS2 skill badge
Real lab and tool task testing arena
Mentorship, real-time support
Coffee, lunch, troubleshooting sessions
🎯 Who This Is For
Engineers serious about high-DOF robotics and AI deployment. Ideal for:
Robotics engineers moving into mobile manipulation
AI/ML practitioners wanting real-world Imitation Learning exposure
Technical founders building service robots or human-assist platforms
PhD students or advanced researchers looking to deepen ROS2 fluency
💻 Prerequisites
Python proficiency + ROS2 fundamentals
Familiarity with GitHub workflows
Some background in ML or motion planning (behavior cloning preferred)
- We offer a free ROS2 prep module — ask us if you need a refresher.
💰 Tuition
$1,995
Includes full hardware access, GitHub repos, certificate, meals, and post-course code support.