Crowd sourced mapping with robust structural features

摘要

A location and mapping service is described that creates a global database of indoor navigation maps through crowd-sourcing and data fusion technologies. The navigation maps consist of a database of geo-referenced, uniquely described features in the multi-dimensional sensor space (e.g., including structural, RF, magnetic, image, acoustic, or other data) that are collected automatically as a tracked mobile device is moved through a building (e.g. a person with a mobile phone or a robot). The feature information can be used to create building models as one or more tracked devices traverse a building.

主权项

1. A computer-implemented method for localizing a mobile device at a structure and generating or updating maps representing the structure, the method comprising:
detect a structural feature associated with the structure using sensor data associated with the mobile device while traversing the detected structural feature; upon detecting the structural feature, obtain signal data associated with the detected structural feature's environment; generate a mobile structure map representing the structure at the mobile device using the sensor data; communicate a feature descriptor to a server communicatively coupled to a dataset comprised of previously received feature descriptors and one or more mobile structure maps generated using the previously received feature descriptors, each feature descriptor associated with a corresponding structural feature and including sensor data used by other mobile devices to detect the corresponding structural feature, signal data obtained by the other mobile devices upon detecting the corresponding structural feature, or a combination thereof; and in response to the communicated feature descriptor, receive a map correction from the server that corrects the detected structural feature's location on the mobile structure map.