Thoughtful Robots

Thanks to “big data” and the ever-increasing speed at which computers are able to process information, machine learning has expanded to the point where robots can interact with humans in ways that were never before possible. Instead of needing to be programmed for specific tasks, machine learning allows them to “think” on their own by identifying patterns based on vast amounts of information.

For example, the recently-developed humanoid service robot EMIEW2 from Hitachi can identify a word – such as “camera” – using speech recognition, compare it to a database of 100 million images, and then locate the object with its built-in imaging sensors. Facial recognition technology also allows it to distinguish between the various people with whom it interacts.

The data that EMIEW2 uses for learning are stored on remote servers and accessed through a wireless network, which allows individual robots to remain compact (80 cm tall) and portable (14 kg). To make it even more user-friendly, EMIEW2 is also designed to travel at the same speed as humans and maneuver on even surfaces (like elevator thresholds) with ease.