Real-World AI Solutions the Focus of Deep Learning for Robotics Summit

Real-World AI Solutions the Focus of Deep Learning for Robotics Summit

Despite the hype, real-world AI solutions are emerging, and a recent summit looked at new techniques and case studies, including a robot for aiding the aging, smart homes, and industrial automation. The buzz around artificial intelligence suggests that it will soon enable fully autonomous vehicles, allow robots to easily serve people’s daily needs, and change how work is done. In late June, global leaders converged at RE•WORK’s Deep Learning for Robotics Summit in San Francisco to learn more about deep learning, neural networks, reinforcement learning, and other advanced AI techniques. More than 300 attendees and 60 speakers discussed current research and real-world AI solutions for industrial applications.

Innovative startups shared their case studies and lessons at the Deep Learning for Robotics Summit. Dr. Shay Zweig, head of AI at Intuition Robotics Ltd., discussed how proactive AI agents are used for decision-making. The Ramat-Gan, Israel-based company’s first application is ElliQ, a social robot for the elderly aimed at reducing loneliness and increasing the quality of life.

Zweig explained that Intuition Robotics has combined algorithms from cognitive computing with heuristics and a variety of learning techniques, from simple statistical models to reinforcement learning. As a result, it can apply active learning to get to know users better and create an agent with multiple goals.

Smart homes benefit from AI advances

“Caspar is building the biggest robot ever, and people will live in this robot,” claimed Ashutosh Saxena, founder and CEO of Caspar AI, which has offices in Redwood City, Calif., and Las Vegas.

Caspar is an intelligent, coherent operating system for homes that uses multi-modal deep reinforcement learning, an intelligent voice assistant, and computer vision to understand the environment in which its users live. It’s also designed to understand the users’ individual needs and to contextualize their requests, Saxena said.

For example, residents can use Caspar for verbal requests such as “Caspar, who is at the door?” “Caspar, can you adjust the shades to block the sun?” or even “Caspar, where are my glasses?”

The real-world AI solutions can also learn to be inactive. For instance, if a user is watching a movie, and someone else walks behind him or her, the light will not turn on because the user gave such feedback by turning off the light with the switch previously.