Polaron Robotics

We make frontier lab-grade Physical AI infrastructure available to everyone. Skip directly to training robots instead of debugging hardware or writing drivers.

  • ✓ Hardware
  • ✓ Software Infra
  • ✓ Cloud Robotics

The Polaron Package

We make building in Physical AI as easy as building in software.

🤖

High Quality Hardware

Robust, low-cost physical AI robotics workcells ready to be deployed to the field.

👥

Teleoperation + Services

Log in to your Polaron from anywhere and teleoperate it yourself, or hire someone from our marketplace.

🎮

Hassle-Free Sim

Run simulation in the cloud with the exact same API as the real robot.

💾

Data Collection Tools

Collect data from your robot or your simulators. Automatically and reliably upload it to your cloud.

🏟️

Real Virtual Fleet

Tired of managing robot farms? Let us manage a fleet for you, so you never have to touch hardware.

💻

Programmable API

Includes LeRobot and ROS2 support, from a local PC or over the internet.

Be the First to Know

Sign up for launch news, behind-the-scenes updates, and early-bird offers. No spam, ever.

Questions & Answers

When will Polaron ship?

Polaron will release an alpha version soon. Sign up for our mailing list for more.

Will Polaron provide ML models?

Our intent is that you, the customer, will provide the models. We may provide some examples for running inference with open-source models.

Will Polaron use our data?

Your data belongs to you. We may roll out a marketplace for robotics data, but that would be strictly opt-in.

Will Polaron provide a mobile robot?

Mobile manipulators introduce a great deal of regulatory constraints and safety challenges, and we would recommend against deploying them given the current state of AI. Polaron is focused on non-mobile systems initially for this reason. Polaron will also introduce innovative form-factors that will enable many tasks to be completed by stationary robots that seem to require mobile ones.

What does Polaron Cloud do?

Polaron Cloud is a secure cloud system that allows you to connect to your Polaron Robot from anywhere over the internet. If you do not enable Polaron Cloud, you can always use your robot over the local network.

Will Polaron provide compute?

Polaron expects inference compute to be provided by either cloud inference, or by a local machine provided by the customer shipped with the robot. In the future, Polaron may roll out local inference hardware as part of the product.

How do Polaron Teleoperators work?

Polaron teleoperators are remote human operators who log in to your Polaron Robot and perform tasks for you. These teleoperators are a completely optional, separate paid service and will not connect to your robot without specific authorization. Note that Polaron may record and review the actions of Polaron teleoperators to ensure quality of service. Polaron will ensure those records remain confidential and are deleted in a timely manner. Teleoperation data will not be used for training any AI or ML models by Polaron.

How is my privacy protected?

Only people you authorize can access your Polaron Robot, and your data is only available to you.

What if I don't want to connect my Polaron Robot to the internet?

You can use your Polaron Robot entirely on your local network. However, many features are dependent on the cloud (such as data upload) and will not be accessible.

What developer options does Polaron provide?

Polaron provides a library that can connect to the robot from your PC, either directly or through Polaron Cloud. The library ships with integrations for LeRobot and ROS2. Additionally, the robot features an Ethernet port so you can attach additional compute to the robot and connect directly over Ethernet.

Will the Polaron Robot be capable of autonomous operation?

The Polaron Robot will not ship with working autonomy.

What experience does the Polaron team have?

The Polaron founder is an ex-Willow Garage ROS core developer, a Google PALM-E infrastructure developer (the original Physical AI infrastructure from before it was called Physical AI infrastructure), and also a team lead on one of the first AI-in-the-loop medical robots.