IT Pro is supported by its audience. When you purchase through links on our site, we may earn an affiliate commission. Learn more

Alphabet tackles industrial robot software market

Intrinsic will teach industrial robots to perform complex tasks

Google parent company Alphabet has launched a new company called Intrinsic that aims to make industrial robots easier to program. 

Originally a project within Google's X research group, Intrinsic has graduated as an independent company within Alphabet. That means it will leave X's rapid prototyping environment and seek development partners. 

CEO Wendy Tan-White explained that conventional industrial robots are difficult to program, often taking hundreds of hours of manual programming to perfect repetitive tasks. Some tasks, such as sanding surfaces of different shapes and sizes or plugging in electrical cords, are beyond robots' reach because they require a subtle awareness of environments and surfaces. 

Intrinsic wants robots to perform tasks like these. It has also been working on programming multiple robots to work together on complex tasks. 

Intrinsic uses various techniques, including artificial intelligence (AI), to make these programming tasks easier by 'teaching' robots how to perform them. The company has also experimented with deep learning, which uses more layers in a neural network to refine machine learning outcomes, and reinforcement learning, which uses feedback on robot performance to improve its attempts at a task over time. 

Other algorithms in its toolbox include automated perception and motion planning, and simulation and force control so robots can adjust how much pressure they apply. 

The company wants robots to apply their training to similar tasks without engineers reprogramming them again from scratch. It also hopes that by becoming aware of their surroundings, they will adapt to changing environments rather than restricting themselves to specific fixed conditions. 

The idea is to make robots more affordable and usable, opening the technology to more businesses. This will also allow companies to make products closer to where they're used, making supply chains more efficient, Tan-White said. 

The Intrinsic team spent five years developing the technology inside the X building. 

Google has dabbled in robotics before, acquiring Boston Dynamics in 2013 before folding it into its Replicant robotics division after reorganizing into Alphabet in 2015. Internal rifts and concerns over brand optics compelled it to sell the company to Softbank in 2017. 

Related Resource

IT Pro 20/20: Understanding our complicated relationship with AI

The 16th issue of IT Pro 20/20 looks at the very human problems associated with artificial intelligence

IT Pro 20/20: Understanding our complicated relationship with AIDOWNLOAD NOW

Google acquired other robotics companies in 2013, including Redwood Robotics, which focused on making robotic arms that are simple to program, and robotic camera systems maker Bot & Dolly. 

Intrinsic is the latest initiative to graduate from X, following in the footsteps of autonomous driving startup Waymo and drone delivery company Wing. 

Other projects still in development within Google X include the Everyday Robot program, a separate robotics program researching whether robots can operate in changeable environments. Its team developed a robot that could sort recycling waste, which is a challenging task given the unpredictable contents of recycling bins. 

Featured Resources

Four strategies for building a hybrid workplace that works

All indications are that the future of work is hybrid, if it's not here already

Free webinar

The digital marketer’s guide to contextual insights and trends

How to use contextual intelligence to uncover new insights and inform strategies

Free Download

Ransomware and Microsoft 365 for business

What you need to know about reducing ransomware risk

Free Download

Building a modern strategy for analytics and machine learning success

Turning into business value

Free Download

Recommended

Tackling our world’s hardest problems with machine learning
Whitepaper

Tackling our world’s hardest problems with machine learning

27 Apr 2022
Building a modern strategy for analytics and machine learning success
Whitepaper

Building a modern strategy for analytics and machine learning success

27 Apr 2022
The COO's pocket guide to enterprise-wide intelligent automation
Whitepaper

The COO's pocket guide to enterprise-wide intelligent automation

1 Apr 2022
Sharpen your manufacturing competitive edge
Whitepaper

Sharpen your manufacturing competitive edge

1 Apr 2022

Most Popular

Russian hackers declare war on 10 countries after failed Eurovision DDoS attack
hacking

Russian hackers declare war on 10 countries after failed Eurovision DDoS attack

16 May 2022
Windows Server admins say latest Patch Tuesday broke authentication policies
Server & storage

Windows Server admins say latest Patch Tuesday broke authentication policies

12 May 2022
Microsoft to double salary budget to retain workers
Careers & training

Microsoft to double salary budget to retain workers

17 May 2022