News

The team's "Sketching" model can effectively take many sounds from the world and generate a human-like imitation of them — like a snake’s hiss and an approaching ambulance siren. Their model can also be run in reverse to guess real-world sounds from human vocal imitations (Image: Alex Shipps/MIT CSAIL with visual elements from Pixabay).

Teaching AI to communicate sounds like humans do

MIT Assistant Professor Sara Beery contributed to the new Tree D-fusion system, which can generate a simulation-ready 3D model of a real tree from images such as those found on Google Street View. The system leverages a tree shape generated using species- and environment-specific data to create realistic, lifelike tree models (Credits: Alex Shipps/MIT CSAIL with background image via Pixabay).

Advancing urban tree monitoring with AI-powered digital twins

The MIT researchers developed an AI-powered simulator that generates unlimited, diverse, and realistic training data for robots. The team found that robots trained in this virtual environment called “LucidSim” can seamlessly transfer their skills to the real world, performing at expert levels without additional fine-tuning (Credit: Mike Grimmett/MIT CSAIL).

Can robots learn from machine dreams?

Spotlighted News

Teaching AI to communicate sounds like humans do

Advancing urban tree monitoring with AI-powered digital twins

Can robots learn from machine dreams?