Unlocking creative and economic potential with industrial robotics
01 - Challenge
01 - Challenge

Reimagining how goods are made and who makes them

Today, industrial robots perform a limited set of tasks in a limited set of industries. That’s because they are typically expensive to purchase, set up, and operate, and require specialized knowledge to program and use. Teaching robots to perform new tasks, like welding the body of a car together, can take hundreds of manual programming hours. Certain tasks, like sanding objects of different shapes and sizes, are often too difficult for robots to even attempt because they lack the perception and tactile skills needed to do them. What’s more, industrial robots usually need to be operated in controlled and unchanging environments because they aren’t equipped to understand and respond to what is happening around them.

The investment of time and expertise required to use industrial robots, along with the fact that they can perform a limited number of tasks in just a few environments like factory floors, make them impractical and out of reach for most businesses.

It can take hundreds of hours to set-up and program industrial robots

Just as the personal computing revolution enabled more businesses and people to access personal computers, industrial robotics is on the cusp of a similar shift thanks to several converging trends. The cost of industrial robot hardware is declining, and the availability of low-cost sensors means it’s possible to gather much more information about the robot’s environment. Thanks to advances in computer vision and machine learning, it is also now possible for industrial robots to perceive and respond to their surroundings and learn how to do more dexterous tasks.

Experimenting with sensors to help robots respond to their environment

What if robots could be as easy to use as computers are today?

The Intrinsic team has been developing software and AI tools that use sensor data from a robot’s environment so that it can sense, learn from, and quickly adapt to the real world. The team hopes that by making industrial robots easier to use, they can enable people to make products and build businesses that we can only just begin to imagine.

02 - Journey
02 - Journey

Exploring how to make industrial robots more useful and flexible

The team started their journey at X by asking a question: How might the team use breakthroughs in robotics and AI to help others reimagine what they can do and make with industrial robots? The team brought experience from robotics, film production, design, computer perception, and mechanical design, and shared a passion for building new tools and technologies that can help unleash creativity in others.

Early experiments to coordinate robot tasks

The Intrinsic team has been exploring how techniques like automated perception, deep learning, reinforcement learning, motion planning, force control, and simulation can be combined to make industrial robots more useful and flexible.

Intrinsic’s early tests have shown that it’s possible for an industrial robot to learn how to perform dexterous tasks and to apply what it has learned from one task to another similar task. Intrinsic has also been able to run several successful tests to orchestrate multiple robots working together. These abilities have the potential to radically reduce the time and complexity of manually programming industrial robots and to make them much more useful than they are today.

Testing in the real world

Intrinsic has also tested their software in real-world settings. For example, the team worked with Gramazio Kohler Research at ETH Zurich to assemble wooden pods for one of their latest architectural projects. They have four ceiling-mounted industrial robots at their Robotic Fabrication Lab to help with the assembly, which involves bringing sets of four panels together at the same time to be glued and cured. This complex task raised the challenge of coordinating the motion of all four robots simultaneously.

With Intrinsic’s motion planning and simulation techniques, they are now able to orchestrate the movements of the four industrial robots to assemble the pods, allowing for a lean and efficient fabrication process.

Orchestrating four industrial robots and two gantries to build wooden pods for a sustainable architecture project ​​by Gramazio Kohler Research, ETH Zurich – Video is sped up
03 - Today
03 - Today

Experimenting with partners worldwide

Today, Intrinsic is working with and learning from partners in different industries and countries. The team is looking for innovation partners, AI experts, software engineers, and roboticists who are excited about using industrial robotics to help millions more people and businesses reach their economic and creative potential.