The warehouse of the future is here, and it is AI-powered. We take a look into the Covariant brain, and how it could revolutionise the way we work.
Californian company Covariant was formed in 2017. For the last two years, they have been quietly developing something extraordinary. This isn’t another exciting prototype or concept. Covariant AI is working right now in real-life scenarios and is showing tangible added value.
I spoke earlier today to Peter Chen, CEO and one of the of Covariant founders, to understand more about their tech, and his vision for the future.
The warehouse of the future
Walk into a warehouse, and what do you see? Conveyor belts, packaging, hundreds of thousands of products, and a continual cycle of monotonous movements. Every product you buy online is picked, sorted, packed and shipped. Over, and over, and over again.
With the increasing demands on commerce to ship faster, cheaper and sooner, companies like Obeta, in Germany, have been looking for ways to adapt.
Here is where Covariant has brought innovation to modern industry. Their Covariant AI is an adaptive learning system, which educates itself through experience.
Let’s think about what that means. The big limitation on AI is that it only knows what we teach it. Whilst there is a lot of dramatisation about the ‘dangers‘ of AI the simple fact remains that it only knows what we tell it. Quite literally.
AI is artificial; it doesn’t replace people. Artificial intelligence is not capable of original thought, of theorising or of using artistic license. It simply does as it is told, and yet can do whatever that might be with higher efficiency than a human.
If you think killer drones are scary, remember – they did not weaponise themselves. No, that was people.
Chen explains that the aim of Covariant is not to replace a human workforce but to help them. He says:
We are familiar with highly automated systems like conveyors, which stop us needing to lift and carry heavy stuff. In the same way, commercial systems are augmented by AI, which can perform highly repetitive tasks, and in injury-prone industries
The task of creating a highly skilled robot to perform in a fulfilment or e-commerce warehouse is not a simple one. The sheer volume of product variables can run into the millions in a large company.
This AI doesn’t just have the capacity for an infinite number of responses but is adapted to use robotics to pick up, handle and correctly allocate individual items. Systems recognise the differences between countless products, SKUs and packaging items.
What is seriously clever, is that the AI continues learning through experience. This means it is quicker than ever to introduce AI into a warehousing supply system. And, if the AI has had similar experiences before, it already knows what it is doing.
Chen explains that we can’t quantify how long it would take to train a new AI-powered robotic system to work efficiently in a new warehouse. This is due to two major factors:
- Covariant AI understands what it has seen before. A large proportion of businesses using this tech are in the fashion and apparel industry, so the AI recognises familiar products, and needs minimal training to learn what to do. In a brand new industry, the AI has to educate itself from scratch which clearly will take longer.
- Automation related software is specific to each organisation. The AI needs time to be tuned up to new warehouse automation systems, but again if it has seen them before, it will already be familiar.
What is particularly fascinating for me is the Covariant brain. In a nutshell, there is only one copy of the AI program.
All the copies of it link back to a central hub; the brain. The ecosystem of this AI family means that each system feeds back experiences via cloud-based communication. This means that, once the AI learns something, it will disseminate that information to every version of the system in use.
The program identifies general patterns to understand each experience. This concept of consolidated learning means that the more the AI is used, the more it learns and the better it performs.
Chen confirms that AI, in itself, has a limitless capacity to absorb knowledge. The size of the Covariant brain system is larger numerically than most others we might have heard of. However, the catch is that AI is restricted by the capabilities that it has today.
Of course, how it develops in the future is quite another matter.
The future of AI
I think this is a perfect example of AI being used to improve the modern workplace. There are so many applications of this program that would prevent workers from being exposed to hazardous conditions.
Think extreme temperatures, dangerous heavy industry and markets where operations are carried out at height or handling sharp materials. Covariant could eliminate that risk to people by using robotics to perform high-risk tasks.
I was interested to ask what Chen sees as the future for AI. Having looked at human-computer interfaces recently, I can foresee a future where high levels of automation mean we can work faster and smarter – but not necessarily harder.
This kind of automation could eliminate the extreme stress environments we hear of, where workers are under pressure to achieve incredibly high pick rates and work tirelessly for the minimum wage to fulfil the demand for next day shipping.
Chen confirms that he imagines AI being positioned to helping people. The system is designed to perform highly repetitive tasks that people simply don’t want to do. It has digital as well as physical applications, so is not restricted to manufacturing or commerce industries. The possibilities are, indeed, endless.
An AI system that uses aggregate experience is extremely powerful, and I believe could revolutionise the way we work around the world. What do you think; could you happily work alongside a machine to make your work-life easier?