Robotics and Embodied AI Atlas

A growing reference for understanding embodied AI and humanoid robotics: definitions, equations, practical examples, and connections between ideas.

These are working notes. Entries identify whether sources still need to be added and whether the explanation has been reviewed.

Use Search to find a concept by name or browse the topic guides and alphabetical terminology index below. Backlinks and the graph reveal connections as this knowledge base grows into longer articles.

Back to Venky Govind’s personal website

Browse by topic

• Foundations and representations
• Policies and predictive models
• Learning and data
• Humanoid motion and control
• Generalization and evaluation

Terminology A–Z

• Action chunk — EA-010
• Action Chunking with Transformers (ACT) — EA-011
• Action space — EA-004
• latent action — EA-005
• Affordance — EA-007
• Behavior cloning (BC) — EA-015
• Benchmark — EA-029
• DAgger — EA-017
• Dataset lineage — EA-027
• Diffusion policy — EA-012
• Embodied AI — EA-001
• Embodiment — EA-002
• trajectory — EA-020
• Goal conditioning — EA-006
• Imitation learning (IL) — EA-016
• Impedance control — EA-025
• Mean per-joint position error (MPJPE) — EA-031
• Motion retargeting — EA-022
• Observation space — EA-003
• Out-of-distribution (OOD) evaluation — EA-028
• Policy — EA-008
• Receding-horizon control — EA-014
• Reinforcement learning (RL) — EA-018
• Residual policy — EA-024
• Sim-to-real transfer — EA-026
• Success predicate — EA-030
• Tracking policy — EA-019
• Vision-language-action model (VLA) — EA-009
• Whole-body control (WBC) — EA-021
• loco-manipulation — EA-023
• World model — EA-013

Further reading

• Paper notes
• Articles