Increasing accuracy in workout autodetection systems and methods
Assigneeadidas AG
InventorStephen John Black
Life of this patent
Estimating the term…
Abstract
Devices, systems, and methods can be used including receiving motion data, categorizing the motion data into portions of a minute that indicate activity or a workout, and automatically determining an accurate number of active minutes for an individual. Multiple data streams may be analyzed and de-duplicated, such that an accurate metric may be computed and reported to an individual.
Claims20
A health and fitness monitoring system for calculating active minutes for an individual, comprising:
- a portable electronic device configured to be carried by the individual, the electronic device comprising:
- a motion sensor configured to generate a first motion data stream including time-stamped motion data;
- a processor;
- and
- a memory,
- the processor configured to process the first motion data stream into an active segment data stream indicating on of an active range or activity classification for a given duration,
- the processor further configured to organize the active segment data stream into a time-stamped minute buckets including duration values for a plurality of activity-specific classifications,
- the memory configured to store the time stamped minute buckets in an array, and
- the processor further configured to determine that a portion of the array includes one of the minute buckets that indicates that the individual engaged in a workout based on the duration values for the plurality of activity-specific classifications, and
- the processor further configured to receive a second motion data stream including third party software platform motion data, wherein the system merges or de-duplicates data within the second motion data stream corresponding to data within the first motion data stream.
A health and fitness monitoring method for automatically detecting an activity, comprising:
- receiving a first motion data stream including motion data for an individual;
- receiving a second motion data stream including motion data for the individual;
- processing, via a processor, the first motion data stream into a motion segment data stream;
- organizing, via the processor, the motion segment data stream into minute buckets to be stored in a memory as an array;
- comparing the array to the second motion data stream to determine that the first motion data stream and second motion data stream overlap, to determine unique data in the second motion data stream corresponding to additional active minutes;
- and
- adding the minute buckets from the first motion data stream to the additional active minutes of the second motion data stream, wherein the first motion data stream comprises an operating system based activity classification.
A health and fitness monitoring method for automatically detecting a workout, comprising:
- comparing motion data from a portable electronic device with third party motion data;
- categorizing the overlap of the motion data from the portable electronic device with the third party motion data;
- determining that a minimum time active has elapsed between an indication that an activity has begun and an indication that the activity has ended such that the activity is categorized as a workout based on the overlap of the motion data;
- and
- adding active time within the third party motion data that is not also within the motion data from the portable electronic device.
A health and fitness monitoring method for automatically detecting an activity, comprising:
- receiving a first motion data stream at a front end of a software platform;
- receiving a second motion data stream at the front end of the software platform;
- transmitting the first and second motion data streams to a back end of the software platform to be normalized and de-duplicated;
- and
- transmitting a third data stream to the front end of the software platform representative of the unique summation of active minutes between the first and second motion data streams.
Dependencies are inferred from the claim text.