Generalist Robot Control
Investigates how generalist AI models can control robotic manipulators reliably without task-specific training or demonstrations.
8 papers
Where this stands
The written synthesis of this thread is for subscribers. Subscribe.
Results across this thread
6 reported results from the papers in this thread.
The table is for subscribers. Subscribe to see every reported number side by side.
How this thread developed
Nov 2026 · Apple
Robots controlled by generalist AI outperform specialists without task-specific training
demonstrates that generalist AI models can control robot manipulators more reliably than specialists without task-specific training or demonstrations.
Nov 2026 · Northwestern University, Carnegie Mellon University
Coding agents control simulated robots less reliably under realistic constraints
Tests coding agents on simulated robotics tasks under realistic constraints, revealing that agents struggle to control robots reliably.
3 further papers
Nov 2026 · Amazon, TU Dresden
AI agents built without training trail ordinary programs on games
Tests direct compilation of game rules into neural network weights without training, a form of zero-shot control for game agents on held-out tasks.
1 further paper
Nov 2026 · Amazon, Rice University
Humanoid robots open round doorknobs without real training
Enables zero-shot transfer of manipulation skills from simulation to real humanoid without real training data
4 of 8 papers shown