Machine Learning (ML) is a game changer for vehicle data collection.  Most systems arestill rather static, where a fixedset of signals are collected.  There are OEMS, that have systems, wheredata collection can be tailoredusing a data acquisition plan, either as collection rules or a data flow network.  Even here, the challenge is to bring this flexibility to the customer.  It is much easier to offer a selection of the signals already being collected than provide a system where the customer can select from the thousands of available signals and configure triggers and rates.  Models could be used both to reduce the data thatneeds to be transmitted and help an OEM data customer to provide a plan that provides only the data actually needed and combine different plans for deploymentto a vehicle.This talk will discuss what is necessary to integrate ML with data collection.