CN-121994251-A - Outdoor intelligent path navigation method based on Beidou satellite and augmented reality

CN121994251ACN 121994251 ACN121994251 ACN 121994251ACN-121994251-A

Abstract

The invention relates to the technical field of computers, and discloses an outdoor intelligent path navigation method based on Beidou satellites and augmented reality. The method comprises the steps of fusing Beidou No. three dual-frequency high-precision positioning and binocular vision-inertia data, calibrating vision scale and globally aligning through Beidou results, switching to vision inertia fusion positioning when signals are weak, combining closed loop detection and correction drift, generating a smooth three-dimensional path based on terrain constraint, dynamically rendering guide marks in AR equipment according to user gestures, dynamically fusing multi-source data through extended Kalman filtering, and guaranteeing positioning continuity and spatial consistency. The invention realizes high-robustness, high-precision and low-delay immersive outdoor navigation.

Inventors

ZENG SHENGQUN
CHEN SHUNXI
WU HONGPING
HUANG LONG

Assignees

深圳市两步路信息技术有限公司

Dates

Publication Date: 20260508
Application Date: 20260410

Claims (10)

1. An outdoor intelligent path navigation method based on Beidou satellite and augmented reality is characterized by comprising the following steps: acquiring original pseudo-range, carrier phase and Doppler frequency shift observation data of a user terminal in an outdoor environment through a Beidou No. three global satellite navigation system receiver, and calculating a preliminary positioning result by combining ephemeris information; Extracting and matching characteristic points of the binocular image frame sequence, constructing a local sparse point cloud map, and performing motion pre-integration based on the six-axis inertial measurement unit data to restrict the visual front end pose estimation process; when the signal quality of the Beidou No. three global satellite navigation system is smaller than a preset threshold value, starting a visual inertia fusion positioning mode, carrying out pose deduction by using the calibrated scale factors and the historical track information, and executing closed loop detection and global optimization after signal recovery to correct accumulated errors; acquiring a target waypoint sequence, path type constraint and terrain traffic class information based on a task planning interface, and generating a three-dimensional navigation path conforming to the current position and the gesture of a user; Projecting the three-dimensional navigation path into a visual field coordinate system of augmented reality display equipment worn by a user, and dynamically rendering a virtual guide arrow, a path contour and a key steering prompt identifier according to the current head posture and the visual line direction; In the augmented reality display process, the availability of Beidou signals, the visual tracking stability and the consistency of inertial data are continuously monitored, the weight of a positioning source and the rendering strategy are dynamically adjusted, and the continuity and the spatial accuracy of navigation guidance are ensured.
2. The outdoor intelligent path navigation method based on Beidou satellite and augmented reality according to claim 1, wherein the Beidou three-number global satellite navigation system receiver is provided with a dual-band antenna, supports B1C and B2a frequency point signal reception, and adopts a precise single-point positioning algorithm to perform positioning calculation by combining region enhancement information.
3. The outdoor intelligent path navigation method based on Beidou satellite and augmented reality according to claim 2, wherein the feature point extraction adopts an improved ORB feature detection algorithm, and a fixed number of key point sampling areas are arranged on each layer of an image pyramid.
4. The outdoor intelligent path navigation method based on Beidou satellite and augmented reality according to claim 3, wherein the motion pre-integration process is constructed based on a Liqun algebraic frame, discrete sampling data of an inertial measurement unit are integrated in a local window, and relative pose increment and covariance matrix thereof are output for restraining visual re-projection error items.
5. The outdoor intelligent path navigation method based on Beidou satellite and augmented reality according to claim 4, wherein the scale recovery is characterized in that continuous position points calculated by Beidou positioning are fitted into smooth tracks, real Euclidean distances between adjacent key frames are calculated, scale factors are determined by ratio of normalized coordinate distances obtained by visual triangulation, and the scale factors are corrected online each time the Beidou effective positioning is updated.
6. The outdoor intelligent path navigation method based on Beidou satellites and augmented reality according to claim 5, wherein the signal quality threshold is jointly determined according to a carrier-to-noise ratio and the number of visible satellites, and when the carrier-to-noise ratio of a B1C frequency point is smaller than 35 dB and the number of visible satellites is smaller than 6, the signal is determined to be unavailable, and a visual inertia fusion positioning mode is triggered.
7. The outdoor intelligent path navigation method based on Beidou satellite and augmented reality according to claim 6, wherein the closed loop detection adopts a dual mechanism of bag-of-words model and geometric consistency verification, when loop is detected, a pose graph is built, global optimization is carried out by adopting a Gauss Newton method, and optimization variables comprise all key frame poses and point cloud map point coordinates.
8. The outdoor intelligent path navigation method based on Beidou satellite and augmented reality according to claim 7, wherein the three-dimensional navigation path generation adopts an A star search algorithm to conduct initial path planning on a grid topographic map, and then conducting smoothing processing through spline interpolation to generate a three-dimensional curve; The terrain traffic class information is derived from a preloaded digital elevation model and vegetation coverage index map for excluding non-trafficable areas.
9. The outdoor intelligent path navigation method based on Beidou satellite and augmented reality according to claim 8 is characterized in that a visual field coordinate system of augmented reality display equipment and the head gesture of a user are synchronized in real time through a nine-axis gesture sensor, the nine-axis gesture sensor fuses magnetometer, gyroscope and accelerometer data, a complementary filtering algorithm is adopted to output Euler angles, the rendering position of a virtual guiding element dynamically adjusts transparency and size according to an included angle between the visual line direction of the user and a path tangent vector, and the transparency is reduced when the included angle is larger than 60 degrees so as to enhance prompt strength.
10. The outdoor intelligent path navigation method based on Beidou satellite and augmented reality according to claim 9 is characterized in that the positioning source weight dynamic adjustment strategy is realized based on an extended Kalman filtering framework, a state vector comprises position, speed, gesture and sensor deviation parameters, an observation vector is composed of Beidou positioning results, vision re-projection residual errors and inertia pre-integration residual errors, and a filter automatically distributes gain weights according to covariance matrixes of all observation sources to achieve optimal fusion of multi-source information.

Description

Outdoor intelligent path navigation method based on Beidou satellite and augmented reality Technical Field The invention belongs to the technical field of computers, and particularly relates to an outdoor intelligent path navigation method based on Beidou satellites and augmented reality. Background With the wide application of Beidou satellite navigation system in the global scope, the outdoor path navigation technology based on satellite positioning has become an important support for intelligent transportation, emergency rescue, personal travel and other scenes. However, in complex environments such as urban canyons, dense forest areas or under overhead bridges, satellite signals are susceptible to building shielding, multipath effects and ionospheric disturbances, resulting in frequent loss of lock, dramatic decline in positioning accuracy and even complete failure of the receiver. Meanwhile, although the visual synchronous positioning and map construction (SLAM) technology based on a monocular or binocular camera can provide continuous pose estimation in a local range, the technology is essentially dependent on image feature matching and motion model deduction, has the unavoidable problems of accumulated drift and scale uncertainty, and is difficult to maintain high-precision absolute positioning capability for a long time. Outdoor intelligent path navigation based on augmented reality is used as a front direction for fusing space perception and man-machine interaction, and higher requirements are put on stability, continuity and absolute accuracy of a positioning system. The technology needs to accurately superimpose virtual navigation information on the field of view of a user in real time in a dynamically-changing real environment, and the premise is to acquire reliable six-degree-of-freedom pose data without drift. The traditional scheme generally adopts a loose coupling mode to respectively process satellite positioning and visual odometer output, lacks a deep fusion mechanism at a state estimation level, and cannot effectively utilize the complementary characteristics of the satellite positioning and the visual odometer output. In the prior art, although the anti-interference capability can be improved by singly relying on Beidou three-frequency signals, positioning continuity can still be lost during signal interruption, and the pure vision SLAM is easy to fail under the conditions of texture missing, illumination mutation or rapid movement, and a global coordinate reference cannot be provided. Even if some systems attempt to introduce Inertial Measurement Unit (IMU) assistance, it is still difficult to address the dimensional drift and absolute position calibration issues in long-term operation. In an outdoor complex scene, when satellite signals are recovered briefly, the initial pose constraint is lacked, and the re-acquisition process is long in time consumption and low in success rate. Therefore, an intelligent path navigation method capable of realizing close coupling fusion of Beidou precise observation data and visual SLAM at a state estimation layer is needed, and stable and reliable augmented reality navigation experience is supported by a bidirectional correction mechanism while high-frequency and low-delay continuous pose output is maintained while absolute positioning accuracy is ensured. Disclosure of Invention The invention provides an outdoor intelligent path navigation method based on Beidou satellites and augmented reality, and aims to solve the problems of navigation failure or drift caused by single satellite signal easy to lose lock and precision dip and accumulated errors and scale uncertainty of a pure vision synchronous positioning and map construction system in complex environments such as urban canyons, forest areas and the like. According to the method, an outdoor path navigation system with environment self-adaptability, state continuity and space consistency guarantee is constructed through a high-precision positioning capability and multisource visual inertia fusion sensing mechanism of a deep fusion Beidou three-dimensional global satellite navigation system, and real-time three-dimensional path guidance with high robustness, high precision and low delay is realized. The invention provides an outdoor intelligent path navigation method based on Beidou satellites and augmented reality, which comprises the following steps: acquiring original pseudo-range, carrier phase and Doppler frequency shift observation data of a user terminal in an outdoor environment through a Beidou No. three global satellite navigation system receiver, and calculating a preliminary positioning result by combining ephemeris information; Extracting and matching characteristic points of the binocular image frame sequence, constructing a local sparse point cloud map, and performing motion pre-integration based on the six-axis inertial measurement unit data to restrict the visual front end pose es