MB
Marcus Bell
3 weeks ago
Finally a book that explains the hard parts without hand-waving. The figures alone are worth it.
Free preview available. Sign in and subscribe to unlock the full book.
Vizuara AI Labs · advanced
Modern Robot Learning
Teach a real robot arm to act by imitation.
Build modern robot learning from the ground up: VAEs, conditional VAEs, transformers, and the ACT (Action Chunking with Transformers) policy. Then train it on a real low-cost SO-101 arm with LeRobot and Hugging Face, and touch policy-gradient RL in simulation.
42 capsules145 figures~8 hoursby Dr. Raj Dandekar
Read on your Kindle or e-reader
Download the EPUB and read offline — perfect for the train. Works on Kindle, Kobo, Apple Books & more.
00Foundations of Robot Learning7 capsules
Foundations of Robot Learning — 7 chapters.
01Why Modern Robot Learning?intuitionfree12 min02Classical Robotics and the Equations of Motionmathfree11 min03Where Classical Robotics Breaks Downconcept🔒11 min04Reinforcement Learning for Robotsconcept🔒11 min05Why RL Is Hard in the Real Worlddeep-dive🔒10 min06Imitation Learning as an Alternativeintuition🔒11 min07Multimodal Data, VLMs, and VLAsconcept🔒11 min01Deep Generative Modeling6 capsules
Deep Generative Modeling — 6 chapters.
08Why Generative Models for Roboticsintuition🔒9 min09The VAE: Encoder, Decoder, and the Bottleneckconcept🔒10 min10The Latent Space: Mean and Variancemath🔒11 min11Sampling and the Reparameterization Trickmath🔒11 min12The ELBO: Reconstruction Loss Meets KL Divergencemath🔒11 min13Building a VAE from Scratchcode🔒11 min02Conditional VAEs3 capsules
Conditional VAEs — 3 chapters.
14From VAE to Conditional VAEconcept🔒12 min15The Math of Conditioningmath🔒12 min16Why CVAEs Matter for Robot Policiesintuition🔒12 min03Transformer Architecture from Scratch6 capsules
Transformer Architecture from Scratch — 6 chapters.
17Attention Is All You Need: A Recapconcept🔒11 min18Self-Attention: Queries, Keys, and Valuesmath🔒11 min19The Encoder-Decoder Frameworkconcept🔒12 min20Vision Transformers: Images as Tokensconcept🔒11 min21Fusing VAEs with Transformers for Roboticsdeep-dive🔒12 min22The Transformer in Codecode🔒12 min04The ACT Policy8 capsules
The ACT Policy — 8 chapters.
23Introduction to the ACT Policyconcept🔒10 min24Hardware Innovations Behind ACTconcept🔒10 min25Software Innovations Behind ACTconcept🔒11 min26The Conditional VAE Inside ACTdeep-dive🔒12 min27The ACT Encoderdeep-dive🔒12 min28Action Chunking Explainedintuition🔒11 min29The ACT Decoder and Temporal Ensemblingdeep-dive🔒11 min30ACT Implementation: Q&A and Gotchasdeep-dive🔒12 min05Training on Real Robots8 capsules
Training on Real Robots — 8 chapters.
31Meet the SO-101 Robotproject🔒9 min32Collecting Demonstrationsproject🔒11 min33Training an ACT Policy from Scratchproject🔒9 min34Running Inference on the Armproject🔒11 min35Debugging Real-World Deploymentdeep-dive🔒10 min36SO-101 Demo Dayproject🔒10 min37Evaluating Beyond Success Ratedeep-dive🔒11 min38Putting It All Togetherintuition🔒10 min06Simulation and Policy Gradients4 capsules
Simulation and Policy Gradients — 4 chapters.
39Why Simulate? The ManiSkill Environmentconcept🔒10 min40The Policy Gradient Theoremmath🔒10 min41Proximal Policy Optimization (PPO)math🔒10 min42Implementing PPO in PyTorchcode🔒11 minRatings & reviews
1,097 readersRate this book
NK
Noah Kim
last month
I read the free preview on a whim and ended up finishing the whole thing in two sittings. Superb pacing.
TA
Tomás Alvarez
5 months ago
Best explanation of this topic I have found anywhere. The sidenotes fill in exactly the gaps I had.
JB
Jonas Bergström
8 months ago
Dense but never confusing. Every chapter builds cleanly on the last — you can feel the care in the structure.