January 11th, 2020

Ann Vole

Rodent tracking roving artificial intelligence

I probably posted this stuff earlier in this blog but there may be subtle changes in plans over the last few years. Two land animals have been tested for a relatively complex verbal language with confirmed results. They were both burrowing communal rodents who are active in the day and have a wide variety of predators visible in a flat landscape. The warning calls contained the equivalent to nouns, verbs, adjectives and adverbs within a single call and for one of these two animal types, unique calls were formed for potential dangers they have never seen before that were very similar in unrelated groups suggesting that these units of information can be combined to describe things that do not have an existing call for. These two animal groups are Richardson ground squirrels located in Alberta and 3 species of prairie dogs found in the southern parts of USA. My city of Regina, Saskatchewan, Canada has a lot of wild Richardson Ground Squirrels as persistent pests and they are legal to keep as pets here. Canada has banned the import of prairie dogs and my province has banned the ownership of prairie dogs to protect the unique subspecies of prairie dogs in this province that have been genetically distinct from USA prairie dogs of the same species for about 20 thousand years due to a minor mountain range along the USA's northern border (North Dakota to Montana). In South America is a smaller rodent group of species called degus who live a lot like ground squirrels but genetically very unrelated (closer to porcupines and guinea pigs and even the rabbit order or mammals than the squirrel-like rodent group that has prairie dogs and ground squirrels). One species of degu is popular in the pet trade and has a very wide variety of vocalizations and the ability to learn and imitate the sounds of other rodents. They also retain their instincts to use a wide variety of warning calls in spite of their use as lab animals for 3 decades prior to 3 more decades as pets. Degus are 5 times smaller than ground squirrels so 5 times cheaper to care for. My plan is to make doors and food dispensers and other interactive robotics that the degus will operate with vocalizations using computers to hear the sounds, interpretative analysis of them, and operate the equipment accordingly (open doors, dispense foods). The secondary goal is to provide a better life for the individuals by allowing more freedom of movement while preventing certain individuals from meeting to limit breeding and prevent fighting. Monitoring the individuals for weight, food consumption, level of activity, sleep schedule, water consumption and many other things will also find, isolate, and inform humans of their change in health automatically. This will require the identity of each individual but there are problems, including cost, for all the identification methods that change the individual (attach something, tattoos, notches cut out of ears or feet). I figured weight would be the best identification method that does not make physical changes to their body. This would also measure how much they eat, pee, poop, and lose in hair loss or dehydration. The problem is trying to connect the weight of someone going through a tunnel to the potential individuals in the areas on each side of that weigh tunnel. If I create a unique copy of an artificial intelligence computer program for each individual, that program can physically move with the animal through data connections between individual computers that control these physical robotics (like doors, weigh tunnels, food dispensers, poop and pee gathering and weighing mechanisms below floor grills, air monitoring, movement sensors). These computers will also need audio transducers like piezoelectric units that can work as both a microphone and a speaker and well into the ultrasonic range. Most computer operating systems are not effective for audio recording or live music creation due to working on interrupts... the processor does a task until some other part of the computer needs the processor to do something and sends the processor an interrupt signal so the processor can stop what it is doing and deal with the interrupt. When you are dealing with live sound processing, you can't stop the processes or you mess up the sound signal. You can do parallel processing though if the change in processing threads is extremely regular and the processor is running fast enough that each thread can get enough processing to do the jobs needed to keep up with the changes in the sound wave it is processing. This means each computer needs an audio-friendly operating system. Moving programs around from computer to computer will be best be accomplished by housing the program in a virtual operating system. These physical computers will also need their operating system to host these virtual operating systems to give them access to their microphones and attached robotics and information storage systems. This is my programming challenge involving operating systems, audio processing in the ultrasonic range, voice recognition software bent to deal with communication methods that are not based on consonants and vowels like human speech.