Foundation model research team · SimpleAI

Simple World Lab

Make the World Simple — Build the Foundation Model for the Physical World

We study how robots act, evaluate their behavior, and learn from experience—connecting embodied pre-training with continual improvement in the real world.

Latest work

Latest research.

More from the lab

Agent + VLA

Diagnosing Semantic Handoff Failures

Studying why skills that work in isolation fail when composed into long-horizon robot tasks.

Embodied data

UMI Data Report

High-fidelity capture, synchronized observations, and replay verification for robot-free manipulation data.

Research focus

Act. Evaluate. Improve.

Three connected directions toward robots that learn beyond their initial training.

01 / EXECUTION

Robotic Agent Systems

How can a robot turn an open-ended goal into reliable, verifiable action?

We connect task-level reasoning with VLA skills, studying how agents plan, compose behaviors, and check their effects in the physical world.

Our focus is the transition between skills: tracking state, detecting failed preconditions, and recovering when execution diverges from the plan.

Long-horizon planningSkill compositionVerification & recovery
02 / PRE-TRAINING

Embodied Pre-training

What data and representations make physical skills transfer?

We study vision-language-action and world-action models trained on UMI, robot, and egocentric experience, with the aim of generalizing across tasks and embodiments.

Data quality is part of the learning problem. We investigate how capture fidelity, temporal alignment, coverage, and data mixtures affect real-robot performance as training scales.

VLA & world-action modelsHigh-fidelity dataCross-embodiment transfer
03 / POST-TRAINING

Continual Learning & RL

How can experience improve both a robot’s skills and its ability to learn?

We are building Simple Evolver to learn from success, failure, and human feedback through real-robot reinforcement learning. World and value models help us study action outcomes, evaluate behavior, and guide policy updates.

We explore meta-learning, online adaptation, and recursive self-improvement (RSI): carrying experience into new tasks while retaining earlier capabilities, with gains tested through repeated deployment.

Real-robot RLWorld & value modelsMeta-learning & RSI

Join us

Build robots
that keep learning.

From foundational models and high-fidelity data to learning on real robots, we are looking for researchers, engineers, and research interns. Explore the teams and application details below.