Thursday, October 06, 2022

Tesla, Artificial Intelligence, and Dog Training

Nikola Tesla

More dogs are being killed every year for lack of training than are actually being trained.


Nikola Tesla invented the remote control (and the first military drone) in 1898 when he maneuvered a tiny ship around a pool of water at a Madison Square Garden demonstration. The ship moved in any direction the audience desired, and even flashed its running lights on and off.

Most of the people watching the demonstration had no idea how it was being done, as very few people knew radio waves existed, much less that they could be used to operate switches.


In 1934, the first attempt at an e-collar was made. Though the radio remote control had been invented 36 years earlier, it was still not being used on dog collars, as batteries were still too large for a dog to carry.

Instead of a battery, this e-collar device was direct-wired through the leash, with the charge being administered by a Ford engine coil (generator) carried by the man.  Needless to say, this device did not take off in the marketplace!

The first patent for a true radio-controlled e-collar appears to have been filed in 1955. 




Commercial e-collar rigs do not seem to have appeared until 1968 or so, when Tri-tronics, Jetco, and Sensitronix marketed competing collars in the pages of Field & Stream.


All three of these commercial e-collars were big and not very reliable.

Their single function was to deliver a very hot level of shock to "bust" dogs off of chasing deer, cars, bicycles, and other "off-game" distractions.


"Buster" collars are pretty far removed from modern e-collars which use tone, vibration, and very low levels of electrical stimulation to teach and "proof" certain behaviors.


The short story with e-collars is that we have gone through the same paroxysm with them that we have with so many transformative technologies, from cars and airplanes, to x-rays and wall sockets.

It always starts with a very good idea, which meets flawed prototypes, which are over-enthusiastically sold.

Misapplication and misunderstanding follow, as do attacks from both "old school" and "new school" competitors.

Improved customer education, improved sales presentations, and more technological innovation follow, but fear-mongering and the natural lag of ignorance, abetted by a flawed narrative based on 30-year old information based on 40-year-old technology, slows down adoption.

In the end, as with electricity, airplanes, and computers, the world embraces the new (and now improved!) and no one can remember a time when we crossed the ocean in schooners, lit our homes with candles, or mailed letters with stamps affixed with human spit.

The way of the future is clear in the world of dog training, and a great deal of it is going to be electronics-based.

Robots with artificial intelligence will be engaged in both reward-based and correction-based dog training, complete with food, vibration, tone, electrical stimulation, visual signals, toys, and human-voice cues.

These robots will have perfect timing, be infinitely patient, and will be as consistent as a well-made Swiss watch.

And will the dogs accept them? We already know they will. After all, isn't a Skinner teaching machine just a robot, albeit one without wheels, and very limited abilities?

Dog training robots will be able to train as many as 80 or 90 dogs a day with a single low-skilled human assistant able to handle three or four robots and the dogs assigned to each.

Will these robots show up tomorrow to make professional dog trainers unemployed?

Nope.

But will they show up 20 or 30 years from now?  

Count on it.  

The world of robots, computers, sensors, and artificial intelligence is now entering an exponential growth curve. Batteries, sensors, capacitors, switches, and controllers are smaller and cheaper than ever before.

The massive strides being made in artificial intelligence, when combined with even more precise GPS technology and robotic production of robots, means that a design-and-production tipping point for robotic dog training is not too far into the future.

That could be wonderful for dogs.

Right now, more dogs are being killed every year for lack of training than are actually being trained.

Yes, that's a crime, but it's probably a solvable crime if we can reduce the cost of basic training by taking the high cost of humans -- and their poor timing, inconsistency, and ignorance -- out of the mix.

No comments: