What is Artificial Intelligence?

Flame, the first ever walking robot to imitate humans

A new robot named Flame was developed by Daan Hobbelen, a PhD student at TU Delft in the Netherlands.

This robot is based on the principles of human walking and this is considered to be most advance robot in its kind. Flame can cope to achieve great stability while remaining energy efficient.

In the two groups of walking robot, Flame is categorized in the second group which is designed to imitate human walking in the sense that the robot is executing a controlled fall forward. This type of robot is not very energy consuming and is not that costly.

Flame is 1.3 m in height and is about 15 kg. It has seven motors. It does not fall and can manage to stay stable because of “organ balance” and a sequence of algorithms which calculate where and how far apart the robot’s ‘feet’ should be placed.

The upper body of Flame has PC104 computer which has a real-time Linux kernel (RTAI) running a control loop and can implement torque and/or position control on all actuated joints.

According to Hobbelen, Flame is the first walking robot with electric actuation. This research might further the development of treatment and diagnosis for people who have difficulty in walking.

Click here to read more about Flame, Artificial Intelligent Robot

Scientists from the Max Planck Institute for Biological Cybernetics discovered that we are able to classify an expression much better when it moves naturally.

Scientists from the Max Planck Institute for Biological Cybernetics in Tübingen, Germany learned that we can recognize facial expressions better when it is moving naturally rather than when it is in a static photograph.

There are many explanations for each facial expression. A frown may say: “Please explain that again!". A nod may signify that you understand.

We must see the expression moving for at least 100 milliseconds in order for us to gain the advantage of dynamic information. Our brain is less capable of decoding the facial motion if the video sequence is shorter.

There are series of experiments that these scientists do in order for them to prove their theories. The result of the experiment shows that either pictures or motion alone are what is needed, but it’s the combination of the temporal sequence and the right facial motion to consistently interpret facial expressions.

Dr. Christian Wallraven, co-author of the study said, "Our results also have implications for the area of computer animation, since its goal is to create artificial avatars and facial animations that are able to communicate realistically and believably."

Click here to read more about Facial Expression In Motion, Computer Animation

Matt Bunting, student with his very own Hexapod

"One of the things I wanted to explore was the idea of reinforcement learning using an artificial intelligence... I wanted it to figure out how to walk straight forward on its (hexapod) own”, said Matt Bunting, a senior at the University of Arizona in electrical engineering.

Hexapod was Bunting's final class project on cognitive robotics. Built from spare parts, and to let the robot adaptively learn how to move he used an Intel Atom processor powering an Ubuntu OS computer. With Logitech Quick Cam to watches what happens when movements are initiated, and a changing scene to determine what effect the movements are having.

Intel ordered themselves two copies to take on an international publicity tour for the Atom processor. A couple days after Matt posted a video of his bot on YouTube, Then, the company who provided Matt with the servos, Crust Crawler Robotics, asked Matt to help develop a software for some of their hexapod kits.

Matt's hexapod Artificial intelligence learning can be applied to tasks other than wait, it can relearn how to walk. The robot even has foot contact sensors that can be used for terrain adaptation.

Click here to read more about Hexapod, Artificial Intelligence

Stanford computer scientists say that, with the diverse rules of every game in computers, they are the ideal tools for exploring ideas in artificial intelligence and fresh approaches to programming.

As Michael Genesereth, computer science professor with the Stanford Logic Group, and Nathaniel Love, a computer science doctoral student, in an article on general game playing (GGP), wrote that: "Programs that think better should be able to win more games.”

He added that, the theorem of general game playing is “drastically different” from the past creation of computer programming.

He said that, “the computer just follows a recipe that has been given to it.” AI’s application is limited to this situation because computer never needs to think for itself. According to him, program like IBM’s Deep Blue shows the smarts of the programmer rather than the smarts of the program. Writing program for GGP is similar as trying to educate a child the way how to play a game.

Genesereth said that competitions between GGP programs are an "evaluation technique for intelligent systems.” And, it is feasible to compare the relative intelligence of its system by playing the programs against one another.

Click here to read more about Artificial Intelligence and Game Programming

An example of a process in artificial intelligent system

An Artificial Intelligent System was first known as an expert system some 3000 years ago. It evolved into sophistication and the use of complicated algorithms in order to ultimately respond and learn from their masters.

A concept of machine communication was introduced by Alan Turing, known as the universal Turing machine that mimics other computing machine. The concept was largely accepted and became the basis for the development of an artificial intelligent system that we now see and use these days.

So how does an artificial intelligent system works?

Let us take for example an artificial intelligent robot.

The artificial intelligent system embedded in the robot's brain takes advantage of sensors integrated in cameras or laser devices. The sensors feed information to the robot's brain. Then, the brain checks for any historical data. If there's none, it creates a composite concept to form part of its memory.

Then the brain builds up the present situation from these concepts. Again, the brain checks for any existing response rule in the memory that is appropriate for this situation. It the picks up one of the matched response rules and sends the action part, to the limbs.

At the start, the robots memory is empty of concepts and response rules. But every time the robot "learns" a new set of response rule containing the present situation, the specific action performed and if it was useful for the objectives set for its actions.

Click here to read more about Artificial Intelligent System, How Artificial Intelligent Works