Agreeing on the format of the design matrix

The IT manager has been around AGV-AI for over 10 years and knows where all the existing data is stored. Pert explains that the design matrix is one of the best ways to start training a ML program. In this case:

  • AGV number identifies the vehicle.
  • Start indicates when the AGV left a location (Location column).
  • End indicates when the AGV reached a pier (Pier column).
  • Distance is expressed in the metric system. Pert explains that most space agencies around the world use the metric system.

The IT manager starts off by saying the following:

  • The AGV number is not stored in the mainframe but the local system that manages the AGVs
  • In the mainframe, there is a start time when an AGV picks up its load at the location and an end time when it reaches the pier
  • The location can be obtained in the mainframe as well as the pier
  • No distance is recorded

Pert would like to have access to the data in the AGV's local system to retrieve the AGV number and correlate it with the data in the mainframe.

The IT manager looks at Pert and decides to tell the artificial intelligence expert the simple truth:

  1. Retrieving data from the AGV guiding system is not possible this fiscal year. Those vehicles are expensive and no additional budget can be allocated.
  2. Nobody knows the distance it takes from a location to a pier. As long as the AGVs deliver the proper products on time at the right piers, nobody so far has been interested in distances.
  3. The AGV mission codes in the mainframe are not the same as in the local AGV guiding system, so they cannot be merged into a dataset without development.

The project is slipping away. AGV-AI doesn't have the leisure or budget to invest man days in extracting data that may or may not be used to optimize warehouse distances in its Amazonization process. Pert has to find a solution very quickly before the very busy IT manager drops the whole idea. Pert decides to think of this overnight and come back quickly the next day.

Keeping an AI project alive means moving quickly.