Robots should notice when conditions change—and adjust before small surprises become costly failures.
Bringing robots into the real world.
Why is adapting to the world so natural for animals—and still so difficult for a robot?
Because they never stop learning from the world they inhabit—adjusting to change while carrying their experience forward.
Evolution spent billions of years refining how life learns in the physical world.
At ACI Lab, we’re searching for the learning algorithm that brings that ability to machines.
Our research direction
A world model for robots that keep learning.
Our research focuses on learning during operation—so robots can adapt through their own experience while preserving useful skills.
One adaptive foundation can support different machines, missions, and environments.
A robot should know when to rely on experience, when to look again, and when to learn something new.
SORP · Introduced by ACI Lab
Second Order
Perception Architecture
SORP explores how a robot’s understanding of the world—and of its own actions—can guide what it observes, how it acts, and when it learns.
Continual learning.
Adapt to new experience while keeping useful skills and knowledge available.
Measured progress.
Test each advance through controlled experiments and simulation, with physical robotics as the next proving ground.
A shared foundation.
Develop one coherent learning architecture for different robotic bodies, tasks and environments.
Research in progress. These are the capabilities we aim to establish.
Understand the world.
And yourself within it.
From its first experience, a robot should learn about the world—and what its own actions can change. That understanding should keep growing throughout its working life, carrying useful knowledge and skills forward.
Research updates · Early access · Collaboration