Saying goodbye to hardware and taking advantage of Google’s software will be the focus of Intrinsic’s progress.
These machines may not be as impressive or attractive as the humanoid robots that Google’s early favorites, but overcoming these control problems may lead to a flood of cheap robots in various industrial fields, which will have a significant impact on the labor market. In fact, the use of machine learning to teach robots to manipulate objects without direct supervision has played Google’s advantage and will obviously be the focus of Intrinsic’s progress.
Recently, Google’s parent company Alphabet’s secret research department X announced the establishment of a robotic software and AI company called Intrinsic, focusing on the development of software for industrial robots. However, the details of what Intrinsic is building, or who its customers are, are unclear.
On July 23, the company’s CEO Wendy Tan White discussed the company’s ambitions in a blog, aiming to use the advancing AI technology and low-cost sensors to create a set of tools that make industrial robots more economical, flexible, and easier to use. “Unleash the creative and economic potential of industrial robots for millions of companies, entrepreneurs and developers.”
Intrinsic is developing software and artificial intelligence tools that can help robots perceive, learn, and adapt to the real world through sensor data. These software will be able to work in a variety of different environments and products, from solar panels to cars.
The company’s CEO Wendy Tan White, a British technology entrepreneur and technology investor, was named one of the 50 most inspiring women in the European technology world in 2017, and was named a female business model in the IT industry in 2017 and an entrepreneur in 2011 , She is an advocate for women to enter the field of technology and investment. Served as Vice President of X.
The shortcomings of current industrial robots are very obvious, and they can only perform a limited number of tasks in a limited industry. It is expensive, and the method of teaching robots how to do things has not changed much in the past few decades. Professional programmer can take hundreds of hours hard-coded manually robot to perform specific tasks such as welding two pieces of metal, the electronic enclosure bonded together , welded together or the vehicle body.
As for many smart and delicate tasks, such as inserting plugs, moving wires, or grinding objects of different shapes and sizes, it is still not feasible for robots because they lack the sensors or software needed to understand their physical environment.
More importantly, industrial robots usually need to operate in a controlled and unchanging environment because they do not have the ability to understand and respond to what is happening around them.
In view of these challenges, the X team believes that breakthroughs in robotics and AI technology can be used to reimagine the capabilities of industrial robots.
Over the past five years, they have recruited a team of leading experts in robotics, computer perception, mechanical design and even film production to actively explore how to integrate automatic perception, deep learning, reinforcement learning, motion planning, force control, and simulation Such technologies are combined to make industrial robots more useful and flexible.
For example, in two hours house training team practiced a robot to complete a required hundreds of hours of programming a USB connection tasks. The team arranged multiple robotic arms to assemble an architectural installation and a simple piece of furniture.
Robots in the Intrinsic Lab use machine learning and force control to plug into three different types of power connectors.
Two robots use perception, force control, and multi-robot planning to assemble a simple piece of furniture.
Intrinsic’s motion planning software coordinated four industrial robots to build wooden pods for the Gramazio Kohler Research sustainable construction project at ETH Zurich.
These early tests show that industrial robots can learn how to perform dexterous tasks and apply the knowledge learned from one task to another similar task. Intrinsic is also able to successfully run multiple tests to coordinate multiple robots to work together.
All this implies that Intrinsic’s software has the potential to fundamentally reduce the time, cost and complexity required to use industrial robots, has long-term potential in helping to solve a wider range of problems and increase the diversity of goods, and can be economically and economically. Produced in a sustainable way.
In Intrinsic’s view, the real advantage of these robots may lie in their impact on climate change. If the industrial robot occupies a smaller area, it can be deployed closer to the consumer manufacturer, which will reduce transportation time and emissions caused by delivery.
X has incubated these technologies for five and a half years, and the team includes many people who worked in the robotics company acquired by Google in the past. Now, it’s time to become an independent company and leave the rapid prototyping environment of the moon landing factory to focus on product development and verification of these technologies.
Company CTO Torsten Kroeger, the company is located in Mountain View, California and Munich, Germany.
Google, the most profitable Internet advertising company, has always wanted to diversify its main income. The reorganized X department of the parent company is like a giant dream factory of the future, actively betting on new technologies in the future. “Other bets” include Verily (biomedical), Waymo (autonomous driving), Wing (drone delivery), Loon (floating Internet in the air), etc. Google has also been obsessed with robotics, but the company’s efforts have not been concentrated and have not yet produced any commercially popular products.
Since Google started manufacturing robots in 2013, the company has spent tens of millions of dollars to acquire robots. In about six months, it has acquired seven companies, including Schaft (a Japanese company famous for its bipedal robots), Bot & Dolly (very cool viral video maker) and Boston Dynamics (no introduction required). At the time, Google’s vice president of engineering Andy Rubin (Andy Rubin) was behind the project, and he named it Replicant to pay tribute to Google’s ambitions. In addition, the name also appeared in a science fiction movie called “Blade Runner”.
But for many years, there was no news from Replicant. Google either sold it or closed most of its acquisitions. Boston Dynamics has now been acquired by Hyundai Korea.
Why? Part of the reason is that robots are difficult to manufacture and not very profitable. But it was also because Rubin left the company after being accused of sexual misconduct in 2014 (Google kept this confidential).
Since the failure of Replicant, the company has paid more attention to software issues. In the past few years, Google has been reinventing its plans to focus on robots that are easier to manage and simpler than humanoid machines.
At that time, many other startups were studying how traditional industrial robots could learn by themselves, such as walking through warehouses full of unexpected objects, including research laboratories OpenAI, Kindred, and startup Embodied Intelligence.
Vince nt Vanhoucke, who helped build Google’s brain, later began to lead Google’s new robotics technology. His staff included many engineers and researchers who had worked under Rubin. These machines may not be as impressive or attractive as the early humanoid robots, but overcoming these control problems may lead to a flood of cheap robots in various industrial fields, which will have a significant impact on the labor market.
In fact, using machine learning to teach robots to manipulate objects without direct supervision is equivalent to playing Google’s (software) advantage, and it will obviously be the focus of Intrinsic’s progress.
“In the past few years, our team has been exploring how to give industrial robots the ability to perceive, learn, and automatically make adjustments when completing tasks so that they can work in a wider range of settings and applications.” Tan White blogs Wrote,
“We have been testing software that uses technologies such as automatic perception, deep learning, reinforcement learning, motion planning, simulation, and force control.”
The company is currently looking for partners in the automotive, electronics and healthcare industries. Among them, these partners are already using industrial robots and hope to learn together.
It’s worth noting that this is not Google’s only robotics project at the moment. In 2019, the company announced the “Everyday Robot” project and caused a sensation. The project is located in the technology incubator X, aiming to develop a “universal learning robot”. What happened to this project is still unclear, and the blog is still staying in 2019. Maybe Intrinsic will have better luck to keep Google’s attention.
Posted by:CoinYuppie，Reprinted with attribution to:https://coinyuppie.com/the-vice-president-of-parent-company-x-personally-takes-charge-and-googles-new-company-is-betting-on-ai-software-to-transform-industrial-robots/ Coinyuppie is an open information publishing platform, all information provided is not related to the views and positions of coinyuppie, and does not constitute any investment and financial advice. Users are expected to carefully screen and prevent risks.