In a recent post in Robotics Online, Jeff Burnstein discusses two possible, and equally likely, scenarios for the development of future robotic platforms: the Honda approach, and the GM approach.
While GM has been using robots for years in its factories, a recent collaboration with NASA on the Robonaut program suggest that GM is interested in using robots to advance automatization for assembly purposes.
Despite being a car manufacturer as well, Honda's strategy is more focused on providing advanced robots to assist people in their homes. As Jeff Burnstein points out, both scenarios, despite apparently leading to different application domains, involve robots "living" side by side with humans, helping them at work, or after work. In order to achieve these ambitious goals, major challenges must be overcome.
It will be challenging to manufacture these robots so that they can operate within the same environment where humans do. More importantly, these robots must be able to interact in a smart way with humans. As you take away robots from their stereotyped, well controlled "universe" of a factory floor, and have them deal with unforeseeable environmental conditions (and human moods...), the complexity of the neural system that will need to equip these robots increases exponentially.
Not only will the industry need low-power, portable chips such as the one that will originate from the SyNAPSE project to support such models, but the industry will need to rapidly come up with complex models able to support equally complex perceptual and cognitive functions.
