It is one of the biggest challenges in Robotics to allow them to interact with their surrounding without the aid of voice or clear pre written set of instructions. So when I heard that there have been some recent advances in this subject so I thought I do some specific searches and compile an up-to-date state of art on this matter.

Roboticists who are working on this "literate artificial intelligence" subject think that developing such robots would be relatively simple because computers are already able to turn scanned books into text and quickly work out what language is used and try and translate it into the language they understand. But the main difficulty is to be able to read different fonts and text scanned from different angles/perspective and distinguishes it from other artificial in the surrounding space. Therefore, the challenge is much bigger than what standard OCR software that can read a scanned page can do.

Here I will try and keep my research tasks simple and just look for examples and prototype robots that perhaps have built in dictionary and spell checker so it can interpret text which is not clearly written.

One would like the basic robot to have abilities so for example be used in rescue operations and work out where they are going inside buildings from signs. Please note that there are situations that remote control may not be possible. Like the recent mining accidents in Chile where the miners were a long way down under the ground.

The advanced task we want the Robot to be able to do is ability to read a label on a closed door you can sometimes get a good idea of what can be found behind it. Therefore it allows the Robot to detect things you cannot directly see. However, with the current status of technology the robots are finding it difficult to distinguish what is writing and what are just random shapes. They still cannot read text on curved surfaces or difficult perspectives.

Lets see what is available right now? It is considered that currently the most famous robot in existence is asimo.honda.com and you can see from the site that currently Asimo is a fairly capable Robot as far as standard movement tasks like Walking, Running, Climbing, Descending, Coordinating the body, and Avoiding obstacles. On the Functions such as, pushing a cart and carrying a tray is no simple task, and he can certainly do them. It can also synchronize with humans in tasks such as shaking hand. In the Intelligence category, Charting a route, recognizing moving objects, distinguishing sounds and recognizing faces and gestures is fairly advanced.

So from the YouTube video you can see how advanced it has become, and you can see that for a Robot like Asimo, the ability to also OCR and read signs in a building and door signs would be so helpful. Distinguishing between Gents and Ladies, knowing that this is Managing Director’s Office would give so much more scope for usefulness of such a device. Otherwise, theoretically one has to programme the Robot with perfect Map of the location as well as give it GPS and mobile location finder tools that will only work outdoor and not much help indoor.

I came across this article in news.discovery.com about Marge and her Her creators, Ingmar Posner and Paul Newman at the University of Oxford, along with their collaborator Peter Corke at Queensland University of Technology. They hope that Marge, and future versions of Marge, could navigate through the real world using the same words and phrases that humans use. Such ability can take Asimo to a new level and make Asimo fairly independent.

The main issue as said earlier is reading Text. "Text spotting is hard because text is a such a variable thing," said Newman. "It appears in so many guises in so many places, in so many sizes, and of course the real world is full of reflections, occlusions, etc."

To Marge, any change in light condition, size of the font, language of the text, angle of view, curved surfaces all add to complexities and therefore add to the computational task even if the right algorithms could be found to digest all these variables..

To attempt this difficult task, it seems that the researchers at Oxford University did something fairly simple, all they did was to in install advanced text recognition software (technically called Optical Character Recognition, or OCR), complete with spell-checker and dictionary, onto Marge, a small robot on wheels. By using a few new tricks to separate text from, say, sticks or trash, and correcting the image based on a simple spell-check and the word's meaning in the dictionary, Marge can bridge the gap in intuition.

These techniques allows her to read newspapers us and theoretically also read the signs in an office or outdoor and make intelligent decisions just like us in everyday life.

I am going to research this subject a little more and follow some other exciting tools that may come handy in making Asimo more intelligent. For example, hacks on the newly available tools such asKinect. For example see this youtube video where the developer first used a single Kinect to video capture of the room into the pc and later used two to produce a good representation of the room in 3D and all in real-time.

Here is another Kinect Robot Hack discussed in another blog in the Personal Robots Group at MIT.

Comments

Name (required)

Email (required)

Website

Enter the code shown above:

Notify me of followup comments via e-mail

About Scifiwood News Reviews and Blogs

These are various short and long News Articles, Reviews and Blogs by Salar Golestanian and employees of SalarO.com as well as contributors of Scifiwood.com. The subject matter are mixed topics with Pure Science to Science Fiction as well as general topics on Web Trends, Technology, Software Engineering genre, or whatever subject that can affect the convergence of today's technology with Science Fiction in any shape or form. These Blogs and Reviews don't have commercial or corporate aspiration, so they are indeed completely independent views. Some of these entries may be short and just link you to the actual news or site that can expand further on the subject of interest. In Phase II we plan to incorporate some Social Networking applications within the portal.