Autotote Factory Initialization - apps4work/co.a4w GitHub Wiki

An essential element of the objective of the autotote is that it is cheap, which goes hand-in-hand with it being a high volume product.

For the auto tote to be a high volume product the deployment of autototes in a factory has to be as simple as possible.

An autotote has mechanisms that can be used to do this. It knows where it is in the factory it knows when it's about to bump into something and it can locate a Target point if it's near it.

Initially it may know where it is, but it does not know where the factory is or where the targets are. The expensive, hard, way to get that information is for a human to configure it. The cheaper way is to let the autotote use the mechanisms it has to discover the factory layout and the targets.

The infrastructure necessary in a factory to deploy is enough pies in the sky to cover the space, and targets each defined by the small laser beam.

The next problem is providing the system with enough information to be able to route the autototes across the factory, without hit obstacles, and to any defined target.

This can be done by a discovery process. A single Autotote in a special discovery mode can use the existing Hardware to create a map of the factory.

Left alone and prevented from escaping the factory the discovery Auto toad would move around the factory essentially at random, and probably rather slowly, encountering and mapping out areas that it can't go, and spotting and remembering the position of targets.

The PIs in the Sky would spot the discovery Autotote and align their sections of the factory using the points where the autototes appears in more than one PI in the Sky's view.

The results of the discovery would be a map that could be presented to a user who would label the targets. She might also declare zones that are physically open to the autototes to be forbidden, and perhaps give hints tohe traffic management about where autototes ought and ought not to go.

Customisation of the system should be limited to the process that says when and where a tote should be taken.

There are plenty of opportunities for new algorithms. For example the definition of a distribution center is that it takes in packages remixes them and send them out.

The process for storing incoming packages it could be as simple as take it from the incoming point and put it somewhere. The process for retrieving them is go get it from wherever you put it and take it to the processing area. The intermediate placement positions are irrelevant as long as the system remembers. And the system can remember a lot more complicated or arbitrary placements then a system that requires humans to fetch them.

In the case of a factory similar to OnPointManufacturing, where the work is taking through a series of steps, it implies some custom application equivalent to the OPM Software Structure, that is basically saying "I done this step on this thing what should I do next?"