Research

Questions worth
engineering around.

Explore the problems, methods, and validation boundaries behind our work.

Embodied AIActive research

Adapting pretrained motion structure to the joint topology, observations, and control cycle of the Unitree G1.

FSQLatent priorsUnitree G1

SystemsActive research

Testing what survives a change in simulator, operating conditions, and inference runtime.

IsaacLabMuJoCoONNX

Sequential learningResearch foundation

A quantitative research stack that shaped our approach to uncertainty, reward design, and honest evaluation.

DQNSequence modelsBacktesting

Evaluation philosophy

Make the assumptions visible.

A model’s output is only meaningful within the conditions used to train, test, and execute it.

Compare additional complexity against simpler alternatives.

Distinguish prediction, decision, execution, and integration errors.

Challenge behavior under noise, uncertainty, and distribution shift.

Connect observations, outputs, commands, and measured responses.

Start a conversation

Discuss a research collaboration, explore a training pipeline, or request a walkthrough of our work.