Brown University researchers have developed an algorithm that helps robots get better at fetching objects. The new algorithm enables robots to actually ask questions when they’re confused, which makes them more intelligent and better at understanding and helping people.
The new research comes from Brown’s Humans to Robots Lab, and is led by computer science professor Stefanie Tellex whose work focuses on human-robot collaboration. Tellex explains that fetching objects is one of the most important tasks for future robot assistants. The problem with robots and fetching objects comes from the fact that the machines get easily confused – they misunderstand and make errors.
So how do we make them more intelligent? Through the use of helping algorithms, of course.
The new algorithm enables robots to receive speech commands and information from human gestures. And when a robot misunderstands, or simply cannot decide which object a person wants, this algorithm enables it to ask for a clarification. It’s worth mentioning that the system works intelligently, which means that the robot doesn’t ask a question with every interaction – only when it’s ambiguous.
The robot and its algorithm were tested, and the trials showed that asking questions intelligently results in better accuracy and speed.
Tellex and her team now plan to combine the new algorithm with robust speech recognition systems, which should further improve the system’s accuracy and speed.
Reference:
Brown University (https://news.brown.edu/articles/2017/03/fetch)
The Wonderful World of Algorithms 