Virtual assistants have been taught to catch thoughts on the fly

logo11d 4 1

Virtual assistants have been taught to catch thoughts on the fly.

The bottleneck in building automation systems is not design and installation, but management.

Design and installation are done by specialists.

They don’t need to be taught, they know what to connect to what and where the signal comes from.

And it is the average person who has to manage all these technological wonders, for whom the remote controls and menus cause fear and horror. If it were simple – you just said it, and it worked as it should, then everything would be much simpler with the advent of automation in our homes. But it is heading in that direction, that soon remote controls will become a thing of the past, like cassette recorders. This is confirmed by the article by Olga Baklitskaya-Kameneva on the website vokrugsveta.ru

Virtual assistants have been taught to catch thoughts on the fly.

“It is not our business to produce gadgets, we only create new technologies,” the head of the German Research Center for Artificial Intelligence (DFKI, Das Deutsche Forschungszentrum für Künstliche Intelligenz), Professor Wolfgang Wahlster, modestly told European journalists about the achievements of his colleagues.

However, these «gadgets» themselves were also presented to journalists: a store of the future, a «smart» apartment and car, and next-generation computer games. All of these are the latest joint developments of the DFKI and the Max-Planck-Institut für Informatik with partners, which are bringing Germany to the forefront in the field of «computer science».

Both institutes are located on the campus of the University of Saarbrücken (Universität des Saarlandes), which allows its scientists to develop new solutions, implement them in programs, and collaborate with professionals from other fields — physicists, chemists, biologists, linguists and psychologists. And while the Institute of Computer Science is mainly engaged in fundamental research, the DFKI focuses on their application in specific technical devices.

Talk to your technology

Imagine you are rushing to the office by car, making your way through the crowded streets of a waking city and dreaming of starting your workday with a Queen song, and your favorite one at that.

You don’t even have to take your hands off the wheel, just clearly say to the minicomputer: “Play me “I’m Going Slightly Mad”!” and the music will start playing. And if the song isn’t in your CD collection, the computer will quickly fish it out of the “world wide web”. Not bad?

Modern MP3 players can store a huge amount of data, but the capabilities of user interfaces, which should actually guide the consumer through the information “jungle”, are limited.

Scientists from DFKI offer their development for communicating with technology called «BabbleTunes».

It is a multilingual system with combined data input, which processes voice commands almost instantly and allows you to control all the main functions of the MP3 player (naturally, you can also use the touch screen).

We saw how it all works in a demo BMW.

The developers suggested that the system prepare a selection of songs by European rock bands from the Internet, then play one that had been written last year. The system responded almost instantly.

To develop new user interfaces that would significantly simplify communication between humans and technology, a special company, SemVox GmbH, was created at the DFKI in July 2008. Its main goal is to give users the choice of whether to communicate with it by giving orders by voice or by gesturing, using a keyboard, a mouse, or a combination of the above.

The idea behind the technology is to extract technical details of the content from the multimodal flow of information.

This technology will help transform your living room into an interactive infotainment center.

“You can calmly leave for a party in the evening, giving the TV the instruction before leaving: “Record the latest episode of The Simpsons for me tonight,” Jochen Steigner, a developer of interactive interfaces and one of the four founders of SemVox, explained to us its advantages in everyday life.

Virtual Poker

A speech synthesis system, capable of conveying emotions, can be seen in action in a virtual casino: there you will be offered to play poker with computer characters — Sam and Max.

You get 52 poker cards, which differ from ordinary ones by built-in chips — RFID tags.

RFID technology allows the identification of an object using radio signals. It is interesting to watch how the emotional state of the virtual players changes during the game, depending on the development of the situation, they comment on the development of the game in different ways with synthetic voices, and, like real players, sometimes bluff.

This is one of the achievements of the IDEAS4Games project, which is being carried out by the DFKI with colleagues from Berlin and Augsburg. It opens up new possibilities for the next generation of computer games.

The main component of IDEAS4Games is the author's developments in the field of artificial intelligence SceneMaker, on the basis of which interactive applications were created.

In order to emphasize the expressiveness of virtual characters in a computer game, the developers used the latest methods of the speech synthesis system and improved modeling of emotions. In combination with the author's developments, a good result is obtained — a flexibly constructed dialogue and improved reactions of virtual characters.

The speech synthesis structure was built on the basis of the MARY (Modular Architecture for Research on speech sYnthesis) system, originally developed by researchers from the DFKI (Language Technology lab) in collaboration with the Institute of Phonetics at the University of Saarbrücken, and to convey the emotional mood of the virtual characters, the scientists used the ALMA (A Layered Model of Affect) computer model.

The emotions and mood of the players are calculated in real time based on the events of the game and in accordance with the 34 rules of poker.

«Our goal is to create virtual characters that can fully participate in a conversation with the user, that is, react not only to the content of phrases, but also to the facial expression of the interlocutor, the timbre and intonation of the voice, behavior, and, ultimately, be able to figure out when a person is smiling and when he is angry.

To this end, together with our European colleagues, we are developing a “Sensitive Artificial Listener” (SAL) system as part of the “Sustained Emotionally Coloured Machine-human Interaction using Nonverbal Expression” (SEMAINE) project. We are already close to teaching a computer to conduct a normal conversation,” says project participant Marc Schröder.

Flying over Google Earth

«Would you like to fly from Saarbrücken to Moscow in a few minutes, straight to your home, right now?» ask the cute guys at the Google Earth screen. Matthieu Deru and Simon Bergweiler from the Advanced Tangible Interface Lab (AdvanTI–Lab) at the DFKI are champions in viewing the institute's videos on YouTube.

The young researchers used one of the latest developments from the gaming company Nintendo, designed for fitness and video games, the so-called Wii Balance Board.

An inexpensive input device that is easy to use is a small platform with built-in motion sensors.

They track how a person changes their position — leans forward, backward, right or left — and transmits it to a computer with special software installed. You stand on the board, look at the screen with a Google Earth map and, changing the position of the body (like an avid surfer), head towards your home, flying over Europe.

By pressing the buttons, you can change the flight altitude. And a fan of computer games in stereo glasses in front of a hemispherical screen will feel like a participant in a virtual plot.

It should be noted that not only museums and computer game creators have shown great interest in the brilliant development of scientists.

In hospitals, the device can be used to rehabilitate patients with vestibular disorders. «We have no doubt that many scientists will follow our example, pick up and develop our ideas,» the guys are sure.

Beethoven and billions of bytes: a search service in the world of music

Forgot the name of the song that's been in your head since morning? Or the author of the melody that you heard?

Whistle or hum it to your computer, and it will give you everything it «knows»: the author, title, all the performers of the works that are in its music database. This amazing search engine is being developed by Meinard Müller, a senior researcher at the Institute of Informatics.

How can you find the music discs you need in a database using just a few bars — a small audio fragment?

Is it possible to compare different performances of a piece of music, synchronize them in time?

How to automatically process the structure of a piece of music on a computer? These questions are of interest to both a music lover and a professional musician.

«Mathematics gives us the tools we need to work out how to solve these problems and use them for specific melodies. My previous research in Michael Clausen's group at the University of Bonn in the field of audio signal processing, the theory of algorithmic representations, algebraic complexity theory and Fourier transforms helps me with this,» says Müller. German conservatories are interested in unusual developments — a good help in classes on music literature.

The store of the future

A «smart» grocery cart will help you choose the right product from the list you have compiled, take you to the right shelf and provide all the information about the purchase.

A digital sommelier will tell you about the flavor bouquet and subtleties of producing alcoholic beverages, compare wines from different vintages and advise which dish is best to serve it with.

The store's refrigerators will store information about the «history» of products, tell you when they arrived in the store, how long they can be stored and under what conditions.

And the products themselves can tell you a lot about themselves: where and when they were made, in what warehouse they were stored, in what combination they are especially tasty, on what assortment a discount is announced.

And, of course, you will have a virtual assistant at your disposal who will explain which products you may be allergic to, and which contain GMOs, preservatives and dyes.

You will soon be able to take advantage of these innovations in the stores of the German retail chain GLOBUS SB. In cooperation with its representatives, the Innovative Retail Laboratory (IRL) was created at DFKI.

Scientists and trade professionals are working on several projects for the store of the future, which will be located on an area of ​​five hundred square meters.

Of course, such a store will need a lot of different electronics — touch screens, RFID tags on products and in the floor (for navigation), antennas on the shelves, temperature sensors… But the main thing is specialized software.

Our acquaintance with «computer science» in German ended for us at the famous Dagstuhl Castle or Leibniz Center for Informatics (Schloss Dagstuhl — Leibniz-Zentrum für Informatik GmbH), which is called the Mecca of informatics: international conferences are held there every month, and the names of all those invited to the castle are entered into the «Dagstuhl Book».

«We and our European colleagues have come out ahead in this area because our research allows us to create working technologies today,» the professor assured me at parting. Reinhard Wilhelm, showing the sights of the famous castle, of which he is the scientific director.

Добавить комментарий

шесть − четыре =

Мы используем cookie-файлы для наилучшего представления нашего сайта. Продолжая использовать этот сайт, вы соглашаетесь с использованием cookie-файлов.
Принять