AUP Teaching Labs

Hands-on Modern AI and Physical AI courses accelerated by AMD GPUs.

Build virtual robots, scale experiments across the GPU, train intelligent policies, and deploy them on physical hardware. The same catalog also takes learners from computer vision and deep-learning foundations to a Tiny LLaMA built from scratch.

64 notebooks 30 Physical AI labs AMD GPU validated

Physical AI

Follow a complete path from simulation and policy learning to real robot manipulation and autonomous navigation.

Genesis language-guided Physical AI agent
Physical Simulation
Genesis Simulation

Advance from robot control and parallel simulation to ROCm vision, tactile perception, and a guarded language-guided agent with an interactive live HUD.

Genesis Simulation
MuJoCo + PyTorch demo
Physical Simulation
MuJoCo + PyTorch

Move from scripted robot control to imitation learning, vision-language-action policies, PPO, and cross-domain reinforcement learning.

MuJoCo + PyTorch
MuJoCo MJX demo
Physical Simulation
MuJoCo MJX

Learn robot simulation with MuJoCo and scale it with JAX, MJX, vectorized rollouts, domain randomization, and Playground PPO.

MuJoCo MJX
Robot Policy Deployment demo
Real Deployment
Robot Policy Deployment

Collect demonstrations on a real SO-101 arm and turn them into deployable ACT and SmolVLA policies.

Robot Policy Deployment
ROS2 Deployment demo
Real Deployment
ROS2 Deployment

Deploy perception, mapping, and autonomous navigation on a real LeKiwi robot using an AMD-only ROCm and ROS2 stack.

ROS2 Deployment

Explore the complete Physical AI pathway →

More AI Courses

Strengthen the perception, modeling, and language foundations behind modern intelligent systems.

Computer Vision teaching lab
Computer Vision

Classification and ResNet, detection, segmentation and SAM, tracking, VAE, and diffusion models.

Computer Vision
Deep Learning teaching lab
Deep Learning

Classical machine learning through neural networks, CNNs, Word2Vec, autoencoders, GANs, and Transformers.

Deep Learning
Large Language Model teaching lab
LLM from Scratch

Autograd and transformer internals through FlashAttention, MoE, LoRA, training, KV cache, and Tiny LLaMA.

LLM from Scratch

Start Learning

Run the notebooks locally with their provided environments. Selected courses also integrate with AUP Learning Cloud for pre-built ROCm-accelerated Jupyter environments.

Acknowledgments

AUP thanks the university partners and open-source communities that make these courses possible.

UniversityProfessors and LabsCourse Contributions
National Taiwan UniversityProf. Chun-Yi Lee, ELSA LabDL, CV
Nanjing UniversityProf. Jingwei Xu, NJUDeepEngineLLM
National Yang Ming Chiao Tung UniversityProf. Ping-Chun Hsieh, Reinforcement Learning and Bandits LabPhysical AI, Reinforcement Learning on MuJoCo

We also thank the AMD AECG team for contributing portions of the Physical AI teaching materials, as well as the Genesis and MuJoCo open-source communities. Detailed source attributions and links to the original repositories are provided in each relevant notebook.