-A Robotic Copilot That Really Understands.

Posted: October 8, 2008 in 2008, Articles
Tags:

Strategy Page:

October 8, 2008: An American firm (Conversay) has developed a voice recognition system that not only does not require “training” (repeating dozens of phrases into a microphone so the software can adapt itself to your speech patterns), but can immediately adapt itself to a wide variety of accents. European air forces are installing this system in their new Typhoon fighters. The new software can immediately tell the difference between American, British, German, Italian and Spanish accented English. Even in the U.S. and Britain, there is a wide variation in how English is pronounced. English is the universal language for commercial and military pilots, at least those that have to operate in foreign airports or airbases. For NATO, this allows pilots from many different nations to speak the same language to each other during joint operations. Ironically, the Conversay software will understand all these accents of English better than many of the human pilots. Incomprehensible English accepts are a common complaint of air traffic controllers dealing with pilots for whom English is a second language.

The French air force pioneered the use of voice recognition in the cockpit. In the 1990s, the French introduced such software, and it even took into account voice distortion under stress (including g stress, as when a fast moving aircraft makes a tight turn.) The new Conversay software builds on this work. The U.S. F-35 is also being equipped with another speech recognition system. This is the first time an American aircraft will be using such a system. The French success with this sort of thing has encouraged other nations to go in the same direction.

With this voice recognition, many tasks that previously required a button push, can now be executed with a spoken command. Tests in actual cockpits have demonstrated accuracy of 98%, which is higher than many human crews are capable of when manually flipping a switch or pressing a button. Typical tasks for spoken commands and electronic ears are requests for information on aircraft condition or changing the status of a sensor or weapon system (which can be presented on the see-through computer display built into the visors of many pilot helmets). A typical speech system can recognize hundreds of words, including some in slurred speech common during high stress maneuvers. The spoken commands save the pilot the time required to press a button or flip a switch, or glance sideways to view a display.

What is developing here is the appearance of, in effect, computerized co-pilots. These systems use computers to constantly collect and examine information from the dozens of sensors on board. These sensors range from the familiar fuel gage, to radar and radar warning devices. Often overlooked by civilians are the numerous calculations and decisions pilots must make in flight. For example, on an interception mission, the pilot must decide how best to approach distant enemy aircraft. Radar will usually spot other aircraft long before weapons can be used. There may also be ground based missile systems aiming radars at you. These conditions present several options; should you go after the enemy aircraft with long range missiles? Or speed up and engage with more accurate cannon and short range missiles? You also have to worry about your own fuel situation, and which of your systems might be malfunctioning. An AI (Artificial Intelligence) computers memory contains the experiences of numerous more experienced pilots as well as instant information on the rapidly changing situation. You can ask your electronic assistant what the options are and which one has the best chance of success. The pilot can then make decisions more quickly and accurately. When enemy aircraft are sighted, the electronic assistant can suggest which of the many maneuvers available are likely to work. If the aircraft is damaged, the electronic co-pilot can rapidly report what the new options are. One becomes quite fond of computers once they have saved your bacon a few times. Many of these capabilities are being installed piecemeal, as part of electronic countermeasures or radar systems. And, bit-by-bit, these “thinking systems” are being merged, producing an electronic co-pilot.

There are other uses for this voice recognition. One option is for one human pilot to lead a group aircraft that included one manned aircraft and three UAVs. The human pilot would be the flight leader, and would give orders to the UAVs. The most dangerous jobs, like putting bombs on heavily defended targets, would go to the UAVs. While the UAVs could also be commanded from the ground, or an AWACs, a human pilot on-the-spot would always have a better view of the situation, and be able to make decisions more quickly. That’s something combat pilots are trained to do.

The British Royal Air Force recently ran a successful test of flight control software that allows the pilot of one warplane to control up to four nearby UAVs. The U.S. Navy has been working on a similar system. It’s all in the software. The UAVs must have software that enables them to do a lot of things by themselves, like flying the aircraft effectively, and being able to understand verbal commands.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s