Robotics Machine Learning Research Scientist - Large Behavior Models
Is deze vacature iets voor u?
Mijn cv maken Maak uw cv en ontdek uw matchpercentage met deze functie — en met alle andere.
Volg uw sollicitaties op mobiel De gratis Whileresume-app, op iPhone en Android.
De functie
Line 1: We are seeking a Robotics ML Research Scientist to advance general-purpose robot manipulation using Large Behavior Models (LBMs) that turn sensor data and human requests into action.
Line 2: Develop data-efficient learning algorithms for robust policies across multiple sensing modalities, including proprioception, vision, 3D representations, force, and dense tactile sensing.
Line 3: Scale learning with diverse data sources, from web-scale text, images, and video to high-quality simulations that augment real-world demonstrations.
Line 4: Research and apply reinforcement learning, offline RL, and behavior cloning for manipulation, while exploring test-time computation and continual learning for long-context tasks.
Line 5: Collaborate on code infrastructure, run experiments on simulated and real robots, publish results in top venues, and contribute to open-source software.
Line 6: Aim to improve robustness and few-shot generalization through self-play and sub-optimal data, and develop interactive agents capable of seeking clarification.
Line 2: Develop data-efficient learning algorithms for robust policies across multiple sensing modalities, including proprioception, vision, 3D representations, force, and dense tactile sensing.
Line 3: Scale learning with diverse data sources, from web-scale text, images, and video to high-quality simulations that augment real-world demonstrations.
Line 4: Research and apply reinforcement learning, offline RL, and behavior cloning for manipulation, while exploring test-time computation and continual learning for long-context tasks.
Line 5: Collaborate on code infrastructure, run experiments on simulated and real robots, publish results in top venues, and contribute to open-source software.
Line 6: Aim to improve robustness and few-shot generalization through self-play and sub-optimal data, and develop interactive agents capable of seeking clarification.
Bekijk de volledige vacature
Taken, profiel, vaardigheden en voordelen — maak gratis een account aan.
of
Al een account?
InloggenDeze vacatures zijn misschien iets voor u
Nog geen echt vergelijkbare vacatures — hier zijn de nieuwste.