Carnegie Mellon University has unveiled new research into robotic navigation in partnership with Facebook’s AI research division, referred to as FAIR.
According to arecent press release, Carnegie Mellon and FAIR are using machine learning to help robots navigate their surroundings by giving them the ability to recognize objects in their path. The SemExp System is advanced enough to know the difference between an end table and a kitchen table and to use the information to pinpoint its location.
"Semantic” navigation is designed to make robots more efficient by understanding the context of their surroundings. The system can "think strategically” about where it’s headed, according to CMU machine learning Ph.D. student Devendra Chaplot.
Chaplot explains that traditional navigation systems build maps of "obstacles,” whereas this system applies machine learning to assign their specifics and function. It stops robots from taking a nonsensical route because, he explains, "Common sense says that if you’re looking for a refrigerator, you’d better go to the kitchen.”
The work is supported by DARPA as well as other military research organizations. Eventually, it could create simpler robot-human interactions because robots would be able to understand directions to retrieve an object in a specific room or to approach "the second door on the left.”
