Graduate students and researchers · 12-20 weeks
Research Path
Study the papers and open problems behind world models, VLAs, cross-embodiment learning, and safety.
Direct answer
Who is the Research Path for?
Research Path is designed for graduate students and researchers and is structured as a 12-20 weeks sequence.
Course sequence
Complete the stages in order. Each page includes definitions, prerequisites, paper explainers, and recommended next steps.
- 01Open →
World Models
Learned models that predict environment dynamics and possible future outcomes.
- 02Open →
Reinforcement Learning
Learning behavior by maximizing expected reward through interaction.
- 03Open →
Vision-Language-Action Models
Models that map visual observations and language instructions to robot actions.
- 04Open →
Spatial Intelligence
Models that understand, generate, and reason about persistent 3D spaces rather than isolated images.
- 05Open →
Humanoid Foundation Models
General-purpose policies and model stacks designed for humanoid bodies, dexterous hands, and whole-body mobile manipulation.
- 06Open →
Benchmarks
Standard tasks and metrics used to compare embodied AI systems.
- 07Open →
Safety
Methods for keeping embodied systems reliable, controllable, and compatible with human environments.
- 08Open →
Deployment
Engineering reliable Physical AI systems under latency, compute, power, hardware, and operational constraints.
- 09Open →
NVIDIA Halos for Robotics
NVIDIA Halos is a full-stack functional-safety system for humanoids, industrial robots, and autonomous mobile robots; Halos OS is its operating-system layer, not a general robot-development OS.