[Home] [Headlines] [Latest Articles] [Latest Comments] [Post] [Sign-in] [Mail] [Setup] [Help] [Register]
Status: Not Logged In; Sign In
|
Science/Tech See other Science/Tech Articles Title: Teaching old robots new tricks: Machines swap knowledge about how to complete a task despite being hundreds of miles apart Roboticists trained a research robots to perform a range of simple tasks These included picking up mugs from a table and placing them on bowls The instructions were stored in a central database known as RoboBrain A robot called Baxter 325 miles (523km) used this information to work out how to perform the same task in a different setting Humans, monkeys and even birds are able to share knowledge and skills, and now robots can too. A machine called Baxter has taken instructions of how to complete a simple task from a robot 325 miles (523km) away and used them to carry out the same job. This is significant because Baxter and the training robot were designed differently and Baxter had to 'figure out' how to complete the task in an entirely different setting. By teaching robots to 'learn' from one another it could give machines new capabilities more quickly and help them adapt to unknown situations and tasks. The research robot, called PR2, was built at Cornell University and was trained using an online game called TellMeDave. Neural Network thinks about your #selfie Convolutional Neural Networks are great: they recognize things, places and people in your personal photos, signs, people and lights in self-driving cars, crops, forests and traffic in aerial imagery, various anomalies in medical images and all kinds of other useful things. But once in a while these powerful visual recognition models can also be warped for distraction, fun and amusement. In this fun experiment we're going to do just that: We'll take a powerful, 140-million-parameter state-of-the-art Convolutional Neural Network, feed it 2 million selfies from the internet, and train it to classify good selfies from bad ones. Just because it's easy and because we can. And in the process we might learn how to take better selfies :) (reference) Yeah, I'll do real work. But first, let me tag a #selfie. Convolutional Neural Networks Before we dive in I thought I should briefly The science of selfies: Neural network analyses 2 million... TellMeDave lets volunteers train robots to perform different tasks using everyday language. As the website explained: 'In order for robots to perform tasks in real world, they need to be able to understand our natural language commands. 'While there is a lot of past research that went into the task of language parsing, they often require the instructions to be spelled out in full detail which makes it difficult to use them in real-world situations. The research robot, called PR2, was built at Cornell University and trained using an online game called TellMeDave. TellMeDave lets volunteers train robots to perform different tasks using everyday language Every behaviour these robots complete is stored in a central database called RoboBrain. An example behaviour is pictured. This database is open-source and can be accessed by other robots and roboticists. The Cornell researchers trained PR2 to pick up mugs and place them on top of upturned bowls Every behaviour these robots complete is stored in a central database called RoboBrain. An example behaviour is pictured. This database is open-source and can be accessed by other robots and roboticists. The Cornell researchers trained PR2 to pick up mugs and place them on top of upturned bowls TRAIN A ROBOT WITH TELLMEDAVE The research robot, called PR2, was built at Cornell University and trained using an online game called TellMeDave. TellMeDave lets volunteers train robots to perform different tasks using everyday language. As the website explained: 'In order for robots to perform tasks in real world, they need to be able to understand our natural language commands. 'While there is a lot of past research that went into the task of language parsing, they often require the instructions to be spelled out in full detail which makes it difficult to use them in real-world situations. 'Our goal is to enable a robots to even take an ill-specified instruction as generic as 'Make a cup of coffee' and be able to figure out how to fill a cup with milk or use one if it already has milk etc. depending on how the environment looks.' Every behaviour these robots completes is stored in a central database called RoboBrain which is open-source and can be accessed by other robots and roboticists. The Cornell researchers trained PR2 to pick up mugs and place them on top of upturned bowls. Baxter then accessed the data from RoboBrain and carried out the same task with different mugs, bowls at Brown University. This involved learning to adapt to the size of the mugs, their weight, and the size and position of the bowls. 'It's pointing in an interesting direction,' said Stefanie Tellex, an assistant professor at Brown University. 'When you put a robot in a new situation - and in the real world it happens in every room the robot goes into - you somehow want that same robot to engage in autonomous behaviors.' The aim is to develop robots that can translate information themselves, based on how their specifications compare to other robots. Read more: Robot Culture: Research Bots Share What They’ve Learned | MIT Technology Review Tell Me Dave Post Comment Private Reply Ignore Thread
|
||
|
[Home]
[Headlines]
[Latest Articles]
[Latest Comments]
[Post]
[Sign-in]
[Mail]
[Setup]
[Help]
[Register]
|
||