An astounding discovery has been made in the world of advances in artificial intelligence. The Defense Advance Research Project Agency (DARPA) as funded a project to create an improved machine to fuse together data which would enable it to act more intelligently. The project’s main objective is to find a way to create a new system that would entice machines to identify rules in a system and solve problems using human-like deduction.
Today’s artificial intelligence robots are making unbelievable advances. In 2011, The Pentagon requested funding to develop novel ways to create algorithms that would allow the machines to integrate information and data from targeting radar, drone-acquired video, cell phone intercepts, and other inputs. The main goal of the project was to develop a machine that would identify and deduce basic objects to engage in problem solving in a complex system.
The intelligent machine is created to understand its purpose and use that information to complete tasks. Until now, artificial intelligence has not been very intelligent at all. It is operated by humans that differentiate between useful and irrelevant objects in a system and then acts on them. However, the scientists at DARPA have been figuring out a way to take human intervention out of the equation and allow the artificial intelligence machine to learn from its surroundings.
This would involve creating an algorithm to allow for advanced sensors for the robot to choose between relevant and irrelevant objects and patterns when completing a task. Scientists from Maryland have just created a new computerized system that makes this possible.
The new setup allows artificial intelligence to differentiate between objects of a specific and non-specific use to complete a task. For example, it allows them to understand YouTube videos to learn the instructions for cooking meals. Furthermore, they have the ability to handle things in their environment and are able to decide what to do without any human guidance or program interventions. The findings and end results of this project were presented at the Association for Advancement of Artificial Intelligence in Austin, Texas.
Reza Ghanadan, project manager of the Science Office at DARPA, stated the MSEE program began with focusing on sensing. This involves the notion of perception and the understanding of a visual scenario. It is not only recognizing and identifying objects, but reasoning why they are used and how they can be used. He also stated that the research team has already started the most important step of the project. This is to create an artificial intelligence machine to understand and observe through an action-grammar module which allows the robot to intelligently deduce language and how to act on instructions from humans and sound and audio recordings.
Astonishingly, such artificial intelligence has the ability to retain knowledge and send it to other robots which makes them exceptional towards other sensors. This demonstrates that these DARPA robots have the ability to understand certain situations after observing what is around it. This shows that new strides in artificial intelligence are a success.
By: Alex Lemieux
Picture: Dan Thornton – Flickr License