FTP-1: A Generalist Foundation Tactile Policy Across Tactile Sensors for Contact-Rich Manipulation
THE PROBLEM
This paper focuses on learning. This is the first Modern Robot LearningFoundation modelA large pretrained model that can be adapted to many tasks. for Perception & SensingTactile sensingTouch sensing through fingers, skin, or contact surfaces. that works across 21 different sensors and Core ConceptsRobotA physical system with sensors and actuators that can observe the world and take actions. embodiments—meaning you can pretrain on diverse tactile hardware and transfer learned Manipulation & TasksManipulationUsing a robot arm or hand to move or interact with objects. skills to new sensors you've never seen before, gaining +31% Simulation & Sim-to-RealSuccess rateHow often the robot completes a task correctly. on unseen tactile setups. It solves the cross-sensor Modern Robot LearningGeneralizationThe robot’s ability to work in new situations it has not seen before. problem that's plagued tactile robotics by using morphology-aware token encoders and a shared Transformer, enabling real Modern Robot LearningTransfer learningUsing knowledge from one task, domain, or robot to help with another. for contact-rich tasks like Manipulation & TasksInsertionPlacing one object into another, like plugging in a connector., Manipulation & TasksAssemblyPutting components together in a structured way., and Manipulation & TasksDexterous manipulationHighly precise object handling, usually with fingers or complex contact.. Read the paper by tracking the Core ConceptsTaskThe job the robot is supposed to complete, such as pick-and-place, navigation, or drawer opening. definition, the Core ConceptsRobotA physical system with sensors and actuators that can observe the world and take actions. or data assumptions, and the evidence that supports the claimed improvement.
HOW IT WORKS
Task framing
Core method
Data and supervision
Evaluation evidence
FIGURES
KEY RESULTS
This is the first Modern Robot LearningFoundation modelA large pretrained model that can be adapted to many tasks. for Perception & SensingTactile sensingTouch sensing through fingers, skin, or contact surfaces. that works across 21 different sensors and Core ConceptsRobotA physical system with sensors and actuators that can observe the world and take actions. embodiments—meaning you can pretrain on diverse tactile hardware and transfer learned Manipulation & TasksManipulationUsing a robot arm or hand to move or interact with objects. skills to new sensors you've never seen before, gaining +31% Simulation & Sim-to-RealSuccess rateHow often the robot completes a task correctly. on unseen tactile setups. It solves the cross-sensor Modern Robot LearningGeneralizationThe robot’s ability to work in new situations it has not seen before. problem that's plagued tactile robotics by using morphology-aware token encoders and a shared Transformer, enabling real Modern Robot LearningTransfer learningUsing knowledge from one task, domain, or robot to help with another. for contact-rich tasks like Manipulation & TasksInsertionPlacing one object into another, like plugging in a connector., Manipulation & TasksAssemblyPutting components together in a structured way., and Manipulation & TasksDexterous manipulationHighly precise object handling, usually with fingers or complex contact..
WHY DEVELOPERS SHOULD CARE
This is the first Modern Robot LearningFoundation modelA large pretrained model that can be adapted to many tasks. for Perception & SensingTactile sensingTouch sensing through fingers, skin, or contact surfaces. that works across 21 different sensors and Core ConceptsRobotA physical system with sensors and actuators that can observe the world and take actions. embodiments—meaning you can pretrain on diverse tactile hardware and transfer learned Manipulation & TasksManipulationUsing a robot arm or hand to move or interact with objects. skills to new sensors you've never seen before, gaining +31% Simulation & Sim-to-RealSuccess rateHow often the robot completes a task correctly. on unseen tactile setups. It solves the cross-sensor Modern Robot LearningGeneralizationThe robot’s ability to work in new situations it has not seen before. problem that's plagued tactile robotics by using morphology-aware token encoders and a shared Transformer, enabling real Modern Robot LearningTransfer learningUsing knowledge from one task, domain, or robot to help with another. for contact-rich tasks like Manipulation & TasksInsertionPlacing one object into another, like plugging in a connector., Manipulation & TasksAssemblyPutting components together in a structured way., and Manipulation & TasksDexterous manipulationHighly precise object handling, usually with fingers or complex contact..
LIMITATIONS
The main limitation to check is whether the claimed behavior holds outside the paper's reported setup. That means testing across different Core ConceptsRobotA physical system with sensors and actuators that can observe the world and take actions. embodiments, scenes, objects, and data distributions.
WHAT COMES NEXT
The practical next step is independent reproduction with clear baselines, ablations, and stress tests. For a developer, the useful follow-up is to map the paper's learning assumptions onto a concrete Core ConceptsRobotA physical system with sensors and actuators that can observe the world and take actions. stack, then test the smallest version of the method that could run end to end.