Intrinsic aims to bring advances in artificial intelligence to large scale industrial robots. Even small companies should be able to easily integrate robotic technology into their processes in the future – even if individual activities require complex sensory perception and thus far actually human tasks.
Google’s parent company Alphabet discovered Intrinsic, a company specializing in software for industrial robots. Intrinsic is a spin-off from X, Alphabet’s incubator for particularly promising and often high-risk projects.
According to her own statement, internal boss Wendy Tan White wants to use her new company to “harness the creative and economic potential of industrial robots for millions of companies, founders and developers.”
Tan White writes, “Typical programmers can spend hundreds of hours programming robots to perform specific tasks such as welding two pieces of metal or gluing together electronics packages.”
Complex tasks such as plugging or laying cables are currently not possible for robots because they lack the necessary sensors or software to sense their environment. Interior wants to change that.
Artificial Intelligence is the focus on the internal
Like its affiliate Google, Intrinsic artificial intelligence In the foreground: the company’s robotic software for robots using techniques from the areas of automated perception, deep learning, reinforcement learning, movement planning, simulation and force control.
Tan White writes, “Over the past few years, our team has been researching how to empower industrial robots to recognize, learn, and automatically adapt so that they work in a wide range of environments and applications. can do.”
According to Tan White, the technical foundation for the interior has been laid in the guise of X over the past five and a half years. Now the company is ready to move towards independence.
Two hours of AI training instead of hundreds of hours of programming work
Tan White describes a project in which a robotic AI learns in two hours to reliably connect three different types of power cables to a circuit board with a force sensor. Otherwise this task would have meant “hundreds of hours” of programming work.
Another example shows robot arms that AI has learned through training and uses sensor data to assemble a simple piece of furniture.
Such demanding tasks for robots are not yet realistic or economical to automate, writes Tan White. There are “millions of such examples in the business world”. Intrinsic’s AI software could “radically reduce” the time, cost and complexity of industrial robots to use, help with more problems and increase the diversity of buildable items.
Tan White describes the fact that more companies may employ machines instead of people as an ecological and economic benefit: currently, ten countries produce 70 percent of the world’s goods, which increases emissions for transportation. Is. In addition, there will be 2.1 million vacant jobs in industrial construction in the United States alone in 2030.
In recent years, Google’s own AI department has repeatedly shown projects in which machine learning should improve robot performance in everyday life, for example service and care robot or a. By Dog robots learn to walk just from video. Google also provided one with Robel Benchmark B for a robot trained in simulation Front.
They: blog x
Read more about robots:
Interior: Alphabet gets new AI robot company was last modified: July 23, 2021 By
