As a Cube insider, you'll get access to interesting new trends, what we think about them and how you can benefit.
New Trend: Robots are Learning Independently
Children often learn on their own through trial and error. They learn to walk by first crawling then trying to stand up, stumbling, and then trying again and again, until eventually they learn to walk without falling. Robot learning has come a long way with continued advances in artificial intelligence; robots can also now learn on their own.
Robots will have to learn in similar ways that humans learn in order for them to be functional and productive in environments such as the warehouse or the home. There are several different learning methods that are used, such as supervised learning, which takes an incredible amount of labeled data that is given to computer algorithms. The algorithms take in millions of labeled data and then become aware of what to look for and essentially are taught what to see. However, this type of learning method is complex and takes a large amount of labeled data.
Many leading scientists focused on research in artificial intelligence are paying more attention to learning methods that do not require as much supervision. Dr. Yann LeCun, VP and Chief A.I. scientist at Facebook says that his “money is on self-supervised learning. This is where computer systems ingest huge amounts of unlabeled data and make sense of it all without supervision or reward. He is working on models that learn by observation, accumulating enough background knowledge that some sort of common sense can emerge.”
New Insight: Learning Robots Are Already Here Today
“In certain cases, for robots to learn just like humans, trial and error can lead to the best performance of understanding a task, which is the thought behind reinforcement learning. Here, the robots learned how to complete a task like pushing a block by getting some information that will assist it, like “seeing” where the block is, and what the nearby terrain is like. Then, a robot gets some measurement of how well it’s doing (the “reward”). The more the robot pushes the block, the higher the reward. The robot had to simultaneously balance exploration (maybe asking itself “can I increase my reward by jumping?“) and exploitation (further exploring behaviors that increase the reward).”
In a recent N³ Innovation Demo Day, Mike Hyslop, VP of Business Development at Dexterity explains Dexterity’s approach for Intelligent Robotics through the use of Agnostic Robots, Machine Learning and Model-Based Control.
In this short video, Mr. Hyslop demonstrates how robots can dynamically identify the ideal way to grasp each item after never seeing the objects previously. Through a sense of touch, the robot can understand how much the product weighs, how hard the product is, which allows the robot to become smarter over time and continue to learn.
New Action: 5 Actions You Can Take Now
Innovation teams can help lead corporations through development and use of robotics to solve business challenges and address labor shortages. Here are 5 actions your team can be taking right now.
Learn about the latest in robotic advancements in your industry.
Find the pain-points and bottlenecks in your manual processes and identify topics for robotic application.
Identify the core robotic technologies that will address the challenges, including those that have had time to mature.
Connect with relevant startups that can help you go faster.
Pilot with one or more startups to test and learn what can be applied and deployed into the business.
N³ Innovation can help you innovate to test and deploy robots that learn. To learn more about how external innovation teams can improve your business, Contact us today!
Want to get the Cube sent directly to your inbox? Sign up HERE
N³ Innovation does not endorse the organizations or products mentioned in this newsletter. If you are interested to learn more, contact us today for a full landscape and targeted solutions.
コメント