Normal laptop
Learn the concepts and inspect small datasets without training a large model.
Start withPushT dataset overview →CPU inspection · 31.6 MB · no robotONE PLACE TO START ROBOT LEARNING
EmbodiedData brings the explanation, requirements, verified source, setup path and beginner tutorial together. You should not need to search through ten different websites just to understand where to begin.
CHOOSE WHAT YOU HAVE
Use the path that matches your current hardware and budget.
Learn the concepts and inspect small datasets without training a large model.
Start withPushT dataset overview →CPU inspection · 31.6 MB · no robotUse a hosted notebook when your laptop does not have an NVIDIA GPU.
Start withPushT training guide →Hosted GPU availability can varyTrain larger policies and move from simulation toward multi-task datasets.
Try nextLIBERO benchmark →Simulation · 40 tasks · intermediateInspect compact real-robot demonstrations before collecting your own data.
Try nextSO-100 pick-and-place →Real-world · actions · languageTHE COMPLETE WORKFLOW
Each stage leads directly to the next and keeps the original publisher visible.
Filter by task, difficulty, hardware and environment.
Read a plain-English explanation and simple terminology.
See storage, compute, actions, language and robot requirements.
Open genuine publisher files and examine one sample.
Follow a practical tutorial with transparent limitations.
Move to a more realistic or advanced dataset.
DESIGNED FOR ACCESSIBILITY
The learning path prioritizes simulation, small downloads, free software and hosted compute—useful for students and self-learners across India and anywhere hardware access is limited.
TRUSTED SOURCES, ONE CLEAR PATH
EmbodiedData explains and connects official resources while the original publishers remain authoritative.
RECOMMENDED FIRST PROJECT
Understand the task, inspect the data and launch your first policy-training command in one guided project.
Start the PushT guide →