News

November 1 2006, Hrafn Th. Thórisson

Over the past 5 years, humanoid robotics have taken a giant leap. What was once only available to large corporations, military sponsored research institutes and eccentric billionaires is now available to the public in various shapes and sizes. Models such as the KHR-1, Robonova and Bioloid can walk, run, do flips and dance — giving homebrewers and researchers on a tight budget a wide range of possibilities to experiment with. But while the progress we've made made in robotic hardware is truly amazing, it seems that these humanoids are missing a vital feature: sensors & communicative protocols.

Reminiscent of the original chess craze of intelligent software, it seems that robotics are now being led down a similar yellow brick road: manufacturers are concentrating on a single aspect of robotics (acrobatics and flexibility) while ignoring the incorporation of features required to make their robots feel more like cognitive beings. As a result they are nothing more than a new kind of remote controlled toy (nowadays often connected to your computer with a cable).

With the tidal wave of smartphones and portable computers washing over us, it makes one wonder when bluetooth or WiFi humanoid robots will become standard. After all, it can put a major delay on launching your brilliant humanoid A.I. architecture if you need to figure out how to mount a camera, 23 sensors and a bluetooth module on your robot without disrupting its delicate balance! (How many programmers does it take to mount a camera on a humanoid robot?)

While the increased agility of humanoids is indeed a joyous development, there remains the question of whether mindmakers around the globe wouldn't rather have a MacMini on wheels with an iSight; instead of a def, dumb and blind bot that does somersaults.

Which leaves us with the question: is there anything we Mindmakers can do as a community to simplify the process of giving these humanoid robots sensors and wireless communication capabilities? The OpenAIR protocol is a step in this direction, although more work is clearly needed!

Plen, the skateboarding humanoid Hitec Robotics, creators of Robonova Robo One Usa, Kondo KHR-1 distributors

August 14 2006, Hrafn Th. Thórisson

Hyperactive Technologies, a Pittsburgh based software company, has licensed it's kitchen production management A.I. system to a Zaxby, a chain of fast food restaurants in the Southern states of America. The system, named Hyperactive Bob, is a large-scale system to help manage the restaurant. Bob gathers information about the restaurants status through real-time sale information, feedback from other employees (through touch-screen displays) and vision to count the cars in the parking lot. The data is analyzed and used to predict incoming orders, and consequently instruct the human employees of what foods to cook and how much.

What's interesting about the system is the high degree of integration in the modules that constitute Bob — mixing vision, purchase transactions and an inventory monitoring system can only be described as a fairly unusual combination. Data-analysis and an active memory system work together to accurately produce maximum restaurant efficiency. According to LiveScience.com, Hyperactive Technologies claim that Bob is more accurate than most seasoned employees. Of course Hyperactive Bob is not an AI like we see in nature: Similar to the expert systems of past, it has no knowledge beyond its field of expertise. If you feed it bogus video, for example, or foil its transaction monitoring mechanisms, not only would it not realize that it was being tricked, it could not even begin to understand it as it has no notion of what being tricked means; its output would become meaningless because it doesn't have any common sense understanding of the world.. Nontheless, an interesting effort and very likely a glimpse of things to come.

LiveScience.com Report Hyperactive Technologies Zaxby's Fast-food Restaurants

August 8 2006, Kristinn R. Thórisson

Waseda University has launched an initiative to create a house inhabited by various kinds of robotic technology, with the aim of ”[inventing] a new social system and principle of　technical designs through the close interaction among humankind and different kinds of robots.” The project was launched late 2005 and involves over 30 researchers. Waseda has a long history of building groundbraking robots, dating back to the '70s. Many of them were humanoid in both form and function. This projects breaks that tradition in the level of integration that they are aiming for: Viewing the full human-robot-habitat system as an integrated unit makes the project likely to bring out issues with prior research from Europe and the U.S. that have taken a more myopic view. Getting funding for projects of this magnitude is generally difficult to do; research tradition in the rest of the world would be likely to result in diverting funding sources away from such an ambitious and holistic view towards projects with a more piecemeal approach. The Japanese have therefore two advantages over and above Westerners: They understand the importance of humanoid robots and they also understand that a holistic approach is necessary to come up with good solutions and technologies.

Project Home Page (English)

March 10 2006, Hrafn Th. Thórisson

WowWee keeps coming up with interesting (weird) new designs, the latest is a robot media hub that aims at humanizing (animalizing?) your regular radio. PC Magazine has a short article on this phenomena that WowWee call SCOTY (Smart Companion Operating Technology), a block-like tower that can move it's head(top block), play your music, read you news and stay alert for unwanted guests. The mediahub reportedly has facial- and speech recognition abilities and acts differently depending on the person it's talking to. I think we can safely say that it would be very fun to play around with this thing, like adding a small LCD screen with a face on it's “head”. WowWee estimates they will start distribution in late 2006, selling at $399 USD.

Read the article on PC Magazine online

Feb 11 2006, Hrafn Th. Thórisson

Palo Alto reports that the Stanford A.I. Lab – the team behind the winning car “Stanley” in the famous 2005 DARPA Grand Challenge race (see picture above), is now planning on bigger and better things. In 2007 they intend to have a new car ready to drive itself from San Francisco to Los Angeles at human-standard speeds. According to Sebastian Thrun, director of the Stanford A.I. labs - the team will concentrate on software integration to create a large system capable of accomplishing the task.

Get the full story on Palo Alto Online

Feb 6 06, Hrafn Th. Thórisson

CMLabs release JavaOpenAIR version 1.0.6, which is their Java implementation of the OpenAIR message and routing protocol. This is an important update to 1.0.5 and everyone is urged to update their AIR Plugs in their Java code with this release. For those working on OpenAIR in other languages: The Java implementation is this reference implementation should be copied as closely as possible, even in other programming languages, to make it easier for users to move between programming languages when programming large systems.

Feb 6 06, Hrafn Th. Thórisson

The first batch of heavily-armed robots has been deployed in Iraq as of January 28th, 2006 reports online magazine Gizmag. The robots, which are based on the earlier Talon bomb-disposal robots can carry many types of weapons, including 66mm rocketlaunhcers. As of yet, the robots are controlled by a human via remote using only minimal A.I. to aid the human user.

Read the article on Gizmag online

Feb 6 06, Hrafn Th. Thórisson

A small article on TMCnet reports that the government of Japan plans to launch a new robot initiative in fiscal 2006. The new initiative counts on very broad applications, from factory robots to household assistants.

Read the newsbulletin on TMCnet.

Feb 6 06, Hrafn Th. Thórisson

The Hindu - India's national newspaper reports online a new facial recognition software developed by Indian researchers capable of automated tagging. A quite impressive story included in the article depicts the power of the software:

“In a dramatic illustration of the power of the tool, the co-founders started a search for Shah and his young son. The search threw up one photo where only Shah was seen… until a zoom showed that his child's photo was there in the shadow — in a photo frame on the wall.”

According to the article, the software is to be launched as a free service for the public in february.

Read the article on "The Hindu online"