CHINA TOPIX

11/02/2024 09:38:35 am

Make CT Your Homepage

MIT Researchers Develop Faster Algorithm for Household Robots

Household robot

(Photo : Christine Daniloff and Jose-Luis Olivares/MIT) Researchers created a new vision system for robots.

Researchers at the Massachusetts Institute of Technology (MIT) have developed a new algorithm that helps household robots identify objects clearly in messy environments.

They believe household robots take advantage of their mobility and the relatively static environment to make object recognition easier. They said robots can do this by imaging objects from multiple perspectives before making judgments about their identity.

Like Us on Facebook

MIT researchers demonstrate that a system using an off-the-shelf algorithm to collect different perspectives can recognize four times as many objects as one that uses a single perspective, while reducing the number of misidentifications.

The first algorithm they tried was developed for tracking systems such as radar. It determines whether objects imaged at different times are same, said lead author Lawson Wong, a graduate student in electrical engineering and computer science.

"If you just took the output of looking at it from one viewpoint, there's a lot of stuff that might be missing, or it might be the angle of illumination or something blocking the object that causes a systematic error in the detector," Wong added.

They've presented a new algorithm that's very accurate and, in some cases, is 10 times as fast as the algorithm that uses different perspectives.

They said this new algorithm can enable robots to identify and handle objects more precisely preset around them. They hope this will make robots more useful and efficient at completing different household tasks.

The new algorithm allows robots to gather multiple perspectives of an object. It even allows robots to quickly gather images and then use the information collected to identify the object.

Details of the new algorithm will appear in the upcoming issue of the International Journal of Robotics Research.

Real Time Analytics