By now we have all heard the term AI or Artificial Intelligence used, and we’ve probably heard that big names like Microsoft, Google and Elon Musk are all in the AI game. A name we probably haven’t heard before is Covariant.ai, a developer of AI technology for warehouse robots. In an article from Wired.com by Will Knight, he brings this startup to our attention and explains how what they’re doing is innovative.
Knight begins by introducing how the winners of this year’s Turing Prize Geoffrey Hinton, Yann LeCun, and Yoshua Bengio, as well as other big names, such as Jeff Dean head of AI at Google, Fei-Fei Lin director of the AI Lab at Stanford, and Daniela Rus head of the Computer Science and AI Lab at MIT have all invested in Covariant.ai versus other companies that develop AI for use in warehouse or industrial context.
The reason these well-versed ‘AI Luminaries’ decided to invest in this company is because of they are focused on AI’s learning capabilities. In 2017 when Covariant.ai was founded, they pioneered the application of machine learning to robotics. Now they’re beginning to make a name for themselves by leading the game of teaching warehouse robots how to grab and sort unfamiliar items through Reinforcement Learning and Imitation Learning. Reinforcement Learning boils down to an algorithm that trains itself through trial and error, similar to how animals learn through positive or negative feedback and Imitation Learning is learning by observing and an algorithm that runs simulations based on the information observed and already learned information. Pretty cool.
Reinforcement Learning as well as Imitation Learning used to be considered unrealistic options due to the amount of computer/computing power required. But as technology continues to advance, the once seemingly insurmountable computer capacity and electricity needed to power the AI tech and corresponding programs has become more realistic, and is even showing success in Covariant.ai’s case.
As demand for robots in warehouse and/or industrial settings continues to increase (prediction of 12% increase in robot production between 2020-2022), companies like Covariant.ai’s fancy AI technology may not be needed to efficiently complete warehouse or industrial tasks, but as their AI quite literally continues to learn and grow, I can see a future of automated or AI technology running our planet.