RT-2: from image and sentence to robot action ↗
RT-2 combines image and language knowledge with robotic action output. Semantic cues should be transferable to the manipulation.
Brohan et al. · RT-2MSD / Research / RESEARCH EDITION
Bewegungen als Verteilung statt einzelner Befehl
Diffusion Policy learns action sequences for robotic manipulation. The project page shows experiments with visual observations and various tasks.
Jetzt eintauchen ↓RECHERCHESTAND 12.11.2026 · FAKTEN + EINORDNUNG + SZENARIENRESEARCH NOTE / 2023Research
Robotics & Physical AI
External research · MSD classification
Keine Behauptung eines eigenen Versuchs.
01 / EINORDNUNG
What this finding means.
Grenzen des Nachweises
Success rates apply to the data and structures examined. A different gripper or a new surface changes the task.
Our derivation for a pilot operation
First create demonstrations and repeatability, then test on unknown objects and disturbances.
This derivation is an editorial recommendation from MSD. It is not a further result of the study cited.
Source and version
Chi et al. · Diffusion Policy · Publikations-/Bezugsjahr 2023. Recherchiert am 12.11.2026.
Read primary source ↗External research, not an MSD customer reference. We will only publish our own experiments with their setup, date, measurements and documented failed attempts.
02 / LINKS
Think ahead. Keep checking.
Many robots, one common data room ↗
Open X-Embodiment bundles data sets from different robotic setups. Shared data is intended to facilitate transfer between tasks and platforms.
Open X-Embodiment CollaborationOcto: an open generalist control ↗
Octo provides an open robotics policy that can be adapted for new observations and action spaces.
Octo Model Team