1) Data is collected: demonstrations (teleoperation, human video), interaction experience or simulation data. 2) A learning paradigm is chosen: imitation learning (behavior cloning), reinforcement learning (reward optimization) or self-supervised learning/pretraining. 3) A model/policy (e.g. VLA, VAM, diffusion policy) learns to map observations to robot actions. 4) The policy is evaluated and transferred to the robot (sim-to-real, fine-tuning on robot data), accounting for safety and generalization.
Hand-programming robots is labor-intensive, brittle and does not generalize to new tasks/environments. Robot learning lets robots acquire and generalize skills from data, coping with the complexity and variability of the real world.
A model mapping observations (vision, sensors, state) to robot actions; e.g. a diffusion policy, a VLA or a VAM.
Official
Demonstrations (teleoperation, human/egocentric video), interaction experience or simulation.
Official
The policy-optimization method: imitation learning, reinforcement learning or self-supervised learning.
Official
Transferring the learned policy to a real robot, bridging the sim-to-real gap and ensuring safety.
Official
Policies learned in simulation often fail on a real robot due to differences in dynamics and perception.
Collecting robot data is costly, and RL can be sample-inefficient.
Learned policies may behave unpredictably out of the training distribution, which is risky for physical robots.
Deep learning begins mapping visual observations directly to robot actions, opening the modern robot-learning era.
Robot foundation models (e.g. RT-2) combine web knowledge with robot action, improving generalization.
Large-scale, multi-embodiment datasets and video-action (VAM) models accelerate efficient robot learning.
Time complexity: Zalezna od metody (IL/RL) i skali danych. Space complexity: O(model polityki + bufor danych/doswiadczenia).
Collecting robot data is costly, RL can be sample-inefficient, and policies learned in simulation do not always transfer to reality (sim-to-real gap).
Imitation learning, reinforcement learning, self-supervised learning or hybrids.
Teleoperation, human/egocentric video, simulation.
Target robot platform (arm, humanoid, mobile).
Depends on the chosen method (IL/RL/SSL).
A general learning paradigm, without a single routing mechanism.
Pretraining and imitation learning are highly parallel; the RL loop and control are sequential.
Training policies (especially from vision/video) and simulation are compute-intensive and benefit from GPU accelerators.
