Sign In

Communications of the ACM

ACM TechNews

Facebook Helped Create an AI Scavenger Hunt That Could Lead to the First ­seful Home Robots

View as: Print Mobile App Share:
The virtual agents seeks items for the scavanger hunt.

Researchers at Facebook and the Georgia Institute of Technology developed a scavenger-hunt challenge that requires a virtual agent to seek an item in a simulated home after parsing a natural-language question.

Credit: Eric Winick

Researchers at Facebook and the Georgia Institute of Technology (Georgia Tech) have developed a scavenger hunt that could help artificial intelligence (AI) programs develop common sense.

The challenge asks AI programs to search for objects inside virtual homes filled with simulated coffee tables, couches, lamps, and other everyday things. The contest requires the AI programs to look for something in the simulated homes after being presented with a natural-language question, the answer to which requires an agent to understand the question and then explore the virtual space in search of the relevant object.

The project relies on reinforcement learning, a form of machine learning inspired by animal behavior, as well as imitation learning, a technique that enables algorithms to learn by observation.

The overall goal is to develop an intelligent system that can see, talk, plan, and reason, says Georgia Tech’s Devi Parikh.

From Technology Review
View Full Article


Abstracts Copyright © 2018 Information Inc., Bethesda, Maryland, USA


No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account