CHANNELS

Aerospace

Automotive

Automated & Autonomous Vehicles

Battery & Electrification Technology

Defense

Drone TV

Electronics

Manufacturing & Prototyping

Materials

Medical

Motion Control

RF & Microwave Technology

Robotics & Automation

Sensors & IoT

Test & Measurement

Aerospace
Automotive
Automated & Autonomous Vehicles
Battery & Electrification Technology
Defense
Drone TV
Electronics
Manufacturing & Prototyping
Materials
Medical
Motion Control
RF & Microwave Technology
Robotics & Automation
Sensors & IoT
Test & Measurement
Aerospace & Defense
Search Results

'Vestri' the Robot Imagines How to Perform Tasks

UC Berkeley  researchers have programmed a robot named Vestri to complete tasks like a baby would - by playing with objects and then imagining how to get the task done. Using this technology, called visual foresight, the robots can predict what their cameras will see if they perform a particular sequence of movements. The robotic learning technology enables robots to imagine the future of their actions so they can figure out how to manipulate objects they have never encountered before. In the future, this technology could help autonomous cars anticipate future events on the road and produce more intelligent robotic assistants in homes.

Topics:
Automation Automotive Motion Control Robotics Software Transportation

More From SAE Media Group

    Transcript

    00:00:02 - [Narrator] Babies are adorable when they're playing with toys, right? But as they play, babies are also learning how their bodies move and how to manipulate objects. Berkeley researchers took inspiration from what they call, babies motor babbling, and programmed that kind of learning into a robot. Meet Vestri. Instead of giving this robot all kinds of measurements for every object and task it may encounter,

    00:00:31 `the usual way they're programmed, researchers let Vestri play around with stuff. Getting used to its own movements while learning the look, shape, and feel of the objects. Vestri did its own motor babbling for about a week. Then it was given a specific task; to move an object from one spot to another. Relying solely on what it learned during playtime, it figures out what to do. It watches its own actions through its camera

    00:01:02 and then imagines what it should do next to get the best results. It produces video scenes that haven't really happened yet, but that it thinks will happen if it pushes the object a certain way. And then carries out the one it decides will get the fastest results. As it moves and imagines future scenes, it continually reconsiders what to do next. Kind of like how we drive a car,

    00:01:35 constantly adjusting our course and speed depending on the road conditions. The robot was given numerous unfamiliar objects. Not drastically varied, but enough to stump robots that don't have this new kind of artificial intelligence. Vestri pushed the objects the right way about 90% of the time. And was even able to maneuver them around obstacles. The Berkeley team is working to expand Vestri's repertoire of objects,

    00:02:06 its ability to work in different environments, and giving it new skills, like lifting, so jobs can get done faster. Whether in the future that'll be tidying up a table, building a house, or who knows what.