HMS Core 3D accurate indoor positioning technology to create a "store-level" travel experience

On April 28, 2022, at the launch of Huawei's folding flagship and full-scenario new products, Huawei Mate Xs 2 folding screen mobile phone is equipped with 3D accurate indoor positioning technology provided by HMS Core Location Kit (Location Kit), providing users with "store-level positioning services ". "Location capability. At present, this capability has covered 160+ cities, 1300+ core business districts, 20+ large-scale transportation hubs and other large and medium-sized public buildings in China, creating an indoor precise positioning travel experience for users.

"3D precise indoor positioning technology" technology base

Indoor positioning technology has a wide range of needs in life scenarios such as indoor shopping guides in shopping malls, high-speed rail and airport travel, as well as in industry scenarios such as tourist guides and takeaway riders to perform contracts at stores.

Currently, there are a variety of indoor positioning solutions in the industry, such as hardware deployment solutions based on Bluetooth tags, UWB base stations, and artificial WiFi/geomagnetic solutions. However, these schemes have shortcomings such as small coverage, high cost, and inability to automatically learn and update.

In order to solve these pain points, HMS Core positioning service has finally realized the accurate indoor positioning function of wide coverage, low cost and self-learning by breaking through a number of key technologies, and supports mainstream mobile OS systems such as HarmonyOS through a unified API interface, ensuring Extending to indoors, you can get seamless precise positioning services.

HMS Core positioning service achieves wide coverage, low cost, and self-learning 3D accurate indoor positioning capabilities through the following four key technologies.

AI Inertial Navigation Trajectory Estimation Technology

Multi-source signal factor graph optimization technology

AI floor recognition technology

Probabilistic modeling high-precision positioning technology

3D precise indoor positioning technology overview

AI Inertial Navigation Trajectory Estimation Technology

AI Inertial Navigation Principle

The industry's common PDR estimation model, based on gait detection and step length estimation, is very sensitive to human height, walking speed, and mobile phone attitude switching, resulting in unstable trajectory length and direction estimation in different scenarios. , the error jitter is large. HMS Core positioning service provides AI inertial navigation trajectory estimation technology to solve this problem. Through the AI self-supervised learning method, the technology can achieve the same trajectory and pose estimation accuracy with only 10% of the original calibration data, which not only reduces the dependence on the amount of training data, but also improves the generalization of user behavior and attitude. In addition, Huawei's self-developed RIO network design has further improved the trajectory estimation accuracy by more than 30% (Huawei laboratory test data), which can meet the positioning needs of various users in real life in complex indoor environments.
Multi-source signal factor graph optimization technology

The principle of multi-source signal factor graph optimization technology

The principle of multi-source signal factor graph optimization technology

The multi-source signal factor graph optimization technology provided by HMS Core location service , based on the basic theory of signal propagation model, introduces the method of graph theory to characterize multi-source signals such as 5G, UWB, FTM, WiFi, BLE, RF, etc. The matching problem is transformed into a graph optimization problem, which not only improves the efficiency of big data processing, but also enables the data model to more accurately characterize the signal map. This technology builds the disordered trajectory distribution into an ordered 2D planar topology signal map, and can finally achieve "store-level" location resolution accuracy as the signal coverage density increases.

Demonstration of building an example from disordered trajectories to ordered 2D topology

AI floor recognition technology

The principle of AI floor recognition technology

Indoor positioning also needs to solve the cross-floor positioning problem caused by users moving on different floors, including scenarios such as stairs, elevators, and escalators. HMS Core location service uses AI to learn cross-floor landmark attributes, and automatically sorts different types of stairs. It uses AI floor recognition technology to achieve 3D floor positioning, with an average accuracy rate better than 95% (Huawei laboratory test data)

Demonstration of construction examples from 2D ordered topology to 3D ordered topology

Probabilistic modeling high-precision positioning technology

Probabilistic modeling principle of high-precision positioning technology

The probabilistic modeling high-precision positioning technology provided by HMS Core positioning service , based on the distribution characteristics of the signal propagation model, proposes a Gaussian process to model the signal residual and adaptive learning hyperparameters, and extend the coverage area of the signal trajectory to the area where the trajectory has not passed. . And with the continuous AI self-learning, it is getting closer and closer to the optimal solution of the model estimation, which makes the positioning more accurate.

In the scene where there is an indoor map, the positioning accuracy can be further improved through the map matching algorithm.

Huawei Mate Xs 2 folding screen mobile phone conference site map

The above technology comes from Huawei's 2012 Riemann Lab, and is open to developers through HMS Core Location Kit. Currently, users of HUAWEI Mate Xs 2 folding screen mobile phones can use the AutoNavi map APP to achieve "store-level" accuracy Indoor positioning experience, more Huawei models will support 3D accurate indoor positioning technology in the future.

At the same time, with the overall acceleration of the HMS ecological layout, the development capabilities of HMS Core positioning services , map services , location services and other core applications have also ushered in rapid improvement. In June 2021, the Petal Maps Platform was integrated and launched, and is fully open to developers and partners. . The Petal Maps Platform map service platform currently has 22 rich map service capabilities. By the end of 2021, it has covered developers in more than 200 countries and regions around the world, accumulated 260 million+ POI data, and supports more than 70 map display and search languages. , the number of access developers exceeds 20,000.

For more information about HMS Core location services, please visit the official website of Huawei Developer Alliance - HMS Core Location Services . For related technologies, please refer to the list of papers.

references:

1. AI inertial navigation trajectory estimation technology:
"RIO: Rotation-equivariance supervised learning of robust inertial odometry", 2022 CVPR, Huawei 2012 Riemann Lab.

2. Signal space to distance constraint modeling technology:
"Mining geometric constraints from crowd-sourced radio signals and its application to information fusion and indoor positioning", 2021 IEEE Access, Huawei 2012 Riemann Lab.

3. High-precision positioning technology for probabilistic modeling:
"Measuring Uncertainty in Signal Fingerprinting with Gaussian Processes Going Deep", 2021 IPIN, Huawei 2012 Riemann Lab.

Learn more details>>

Visit the official website of Huawei Developer Alliance
Get development guidance documents
Huawei Mobile Services Open Source Warehouse Address: GitHub , Gitee

Follow us to know the latest technical information of HMS Core for the first time~