Delivering accurate location for people moving about in GPS-denied areas – such as when indoors and underground – has not been realistic to date without the use of costly infrastructure or manual RF mapping of facilities. Delivering seamless and ubiquitous indoor location is difficult, given that GPS is not available, compasses are highly inaccurate, floor plans may not be available, and relying upon accurately mapped infrastructure may not be a viable option. The TRX NEON solution implements sensor fusion and mapping to deliver accurate 2D and 3D feature maps, enabling location where GPS is unavailable, without relying upon pre-installed or networked infrastructure. Core to the TRX technology is the ability to dynamically produce distinct feature maps of indoor spaces, enabling fusion of information from multiple devices to populate maps that include building structural features, elevators, stairwells, and transitions.
Existing Location Technologies
GPS is primarily an outdoor tracking technology, with little or no accuracy in most structures. It is plagued by errors on the order of tens of meters (and often more), even outdoors, when in close proximity of buildings or other obstructions. Wi-Fi can be a valuable assist - making a cell phone aware of the building it is inside. However, without extensive RF mapping (often called "fingerprinting"), Wi-Fi itself can deliver only general location awareness, and such mapping can be highly manual and labor intensive. Even once mapped, WiFi can be susceptible to errors driven by interference, changes to building infrastructure, and reconfiguration of networks. Ultra wideband and tagging are valuable, but only for applications where tracking is desired by a building owner who invests in installing a dense array of devices in the building.
NEON Sensor Fusion and Dynamic Mapping for Location
NEON implements a suite of patented sensor fusion and dynamic mapping algorithms, which intelligently correlate information from a broad range of sensors (magnetic, accelerometer, gyroscope, light, pressure, RF, etc.). The intelligence of sensor fusion portion of the NEON Navigation Engine lies in part in its ability to isolate and select the sparse areas in which a degraded sensor’s (e.g., compass, GPS) estimates are accurate, and eliminate the rest. This distinguishes NEON from filters and systems that severely drift away from the truth in the presence of a majority of incorrect predictions from a sensor.
Neon uses received sensor information to infer site maps, building features, and other landmarks (distinct building and transition features) dynamically as people move about an area or building. Information from multiple people moving about in the same building is merged to deliver crowd-sourced enhancements to inferred maps, delivering accurate features maps that enhance location estimates. Additional ranging information (e.g., from known beacons), if available, can also be used to constrain location results. Such ranging Information includes ranging between people operating close together and between a person and a fixed RF node if infrastructure is available.
Given known maps, feature or floor plan information, accuracy will be further increased as NEON will match location estimates and inferred maps and features to known floor plans and maps, and implement wall avoidance. User initialization and corrections, where available, are also incorporated into the NEON solution.