Glasses guide direction recognition and content triggering system integrating geomagnetism and inertial navigation

Abstract

The invention belongs to the technical field of data processing and analysis, and particularly relates to a geomagnetic and inertial navigation fused glasses guide direction identification and content triggering system, which comprises a sensing compensation component, a calculation storage component, a guide resource component and a display component, the sensing compensation assembly comprises an inertial measurement unit, a geomagnetic sensor, an active magnetic compensation coil and a coil driving module; the calculation storage assembly comprises a control processor and a memory, and the control processor is used for driving the active magnetic compensation coil to obtain a coil compensation geomagnetic vector according to a phase coding zero-returning sequence, executing attitude propagation based on the inertial measurement unit, and executing course updating on the coil compensation geomagnetic vector based on Huber anti-outlier loss to obtain a head attitude. According to the invention, the influence of magnetic disturbance and attitude jitter on direction identification and content triggering can be effectively reduced, and the stability, consistency and practicability of the glasses guide system in a real use scene are improved.

Description

Technical Field

[0001] This invention belongs to the field of data processing and analysis technology, specifically relating to a glasses navigation direction recognition and content triggering system that integrates geomagnetism and inertial navigation. Background Technology

[0002] In existing technologies, inertial measurement units (IMUs) typically consist of gyroscopes and accelerometers, used to estimate head attitude changes. Gyroscopes can provide continuous angular velocity information over a short period, making them suitable for capturing rapid head turns. However, gyroscope outputs exhibit zero bias and random drift, and long-term integration leads to a gradual accumulation of heading errors. Accelerometers provide gravity direction information for correcting pitch and roll, but during walking, climbing, or turning, accelerometers are susceptible to linear acceleration interference, resulting in unstable gravity direction extraction. To suppress gyroscope drift, existing technologies typically incorporate geomagnetic sensors as a heading reference, using a combination of geomagnetic vectors and gravity direction to correct heading.

[0003] However, the reliability of geomagnetic sensors is poor in actual wearing environments. The glasses themselves often contain speaker magnets, drive coils, power circuits, and metal structural components. These components can create localized magnetic disturbances near the geomagnetic sensor, causing the geomagnetic vector to deviate from the actual direction of the Earth's magnetic field. Furthermore, external sources of magnetic disturbance in the user's environment, such as elevators, power facilities, vehicles, and commercial equipment, can cause abrupt changes or long-term shifts in the geomagnetic sensor output. Current technologies commonly employ static calibration of geomagnetic data or simply threshold-removal or weight reduction of abnormal geomagnetic data in the fusion algorithm. These methods are effective when magnetic disturbances are weak, but in scenarios where magnetic disturbances are persistent or frequently changing, geomagnetic observations are either completely discarded, forcing the heading to rely solely on a gyroscope for short periods, or incorrectly fused, leading to heading jumps. Summary of the Invention

[0004] Therefore, the main objective of this invention is to provide a glasses-based navigation direction recognition and content triggering system that integrates geomagnetism and inertial navigation. By introducing an active magnetic compensation coil around the geomagnetic sensor and driving it with a phase-encoded zero-return sequence by a control processor, geomagnetic observation remains under control even in complex wearing environments and under external magnetic disturbances. Combined with attitude propagation from the inertial measurement unit and heading updates based on Huber anti-outlier loss, continuous and stable head posture is obtained. On this basis, the gaze direction is generated using the current position and head posture, and navigation content is triggered by dwell time determination. This ensures that the presentation of navigation information is based on continuous alignment and spatial consistency, thereby effectively reducing the impact of magnetic disturbances and posture jitter on direction recognition and content triggering, and improving the stability, consistency, and practicality of the glasses-based navigation system in real-world usage scenarios.

[0005] The technical solution adopted in this invention is as follows: A glasses-based navigation direction recognition and content triggering system integrating geomagnetism and inertial navigation includes: a sensing compensation component, a computing and storage component, a navigation resource component, and a display component. The sensing compensation component includes an inertial measurement unit, a geomagnetic sensor, an active magnetic compensation coil, and a coil drive module. The computing and storage component includes a control processor and a memory. The control processor drives the active magnetic compensation coil to obtain the coil-compensated geomagnetic vector according to the phase-encoded zero-return sequence, and performs attitude propagation based on the inertial measurement unit and performs heading update on the coil-compensated geomagnetic vector based on Huber anti-outlier loss to obtain the head attitude. The navigation resource component provides the current position and stores the navigation content library. The control processor generates the gaze direction based on the current position and head attitude, performs dwell determination to trigger the corresponding navigation content in the navigation content library, and outputs it through the display component.

[0006] Furthermore, the active magnetic compensation coil includes a first compensation coil segment and a second compensation coil segment disposed on the eyeglass body. The coil driving module includes a constant current driving unit and is electrically connected to the first compensation coil segment and the second compensation coil segment respectively. The constant current driving unit is used to output a constant driving current during the duration of each sampling frame corresponding to the phase-encoded return-to-zero sequence, and to control the driving current ripple within the current ripple constraint range stored in the memory.

[0007] Furthermore, within each sampling period, the control processor organizes the phase-encoded zero-return sequence into a series of sequentially arranged zero-magnetic sampling frames, first excitation sampling frames, second excitation sampling frames, and zero-return sampling frames. The zero-magnetic sampling frame corresponds to the coil drive module outputting a zero-drive current and triggering the geomagnetic sensor to sample and obtain a zero-magnetic vector. The first excitation sampling frame corresponds to the coil drive module outputting a first positive drive current to the first compensation coil segment and triggering the geomagnetic sensor to sample and obtain a first excitation geomagnetic vector. The second excitation sampling frame corresponds to the coil drive module outputting a second positive drive current to the second compensation coil segment and triggering the geomagnetic sensor to sample and obtain a second excitation geomagnetic vector. The zero-return sampling frame corresponds to the coil drive module simultaneously outputting a zero-return drive current to the first compensation coil segment and the second compensation coil segment and triggering the geomagnetic sensor to sample and obtain a zero-return geomagnetic vector.

[0008] Furthermore, the control processor generates a synchronization trigger instruction in each sampling cycle and sends it to the coil drive module and the geomagnetic sensor simultaneously. The synchronization trigger instruction is used to constrain the coil drive module's drive current establishment phase and the geomagnetic sensor's sampling window to be aligned within the same time base, and to constrain the zero magnetic sampling frame, the first excitation sampling frame, the second excitation sampling frame, and the return-to-zero sampling frame to be completed in a fixed frame order.

[0009] Furthermore, the control processor constructs a coil magnetic field response fingerprint based on the zero magnetic field vector, the first excitation magnetic field vector, and the second excitation magnetic field vector. The coil magnetic field response fingerprint includes at least the change in the first excitation magnetic field vector relative to the zero magnetic field vector and the change in the second excitation magnetic field vector relative to the zero magnetic field vector. The change in the first excitation magnetic field vector relative to the zero magnetic field vector is used to characterize the direction and intensity of the equivalent magnetic field exerted by the first compensation coil segment on the geomagnetic sensor. The change in the second excitation magnetic field vector relative to the zero magnetic field vector is used to characterize the direction and intensity of the equivalent magnetic field exerted by the second compensation coil segment on the geomagnetic sensor.

[0010] Furthermore, the control processor extracts the gravity direction based on the acceleration sequence output by the inertial measurement unit, and decomposes the zero-magnetic geomagnetic vector into a vertical component parallel to the gravity direction and a horizontal component orthogonal to the gravity direction. At the same time, the zero-return target geomagnetic vector is decomposed into a target vertical component and a target horizontal component. The control processor generates a zero-return compensation vector based on the difference vector between the target horizontal component and the horizontal component. The zero-return compensation vector is used to constrain the active magnetic compensation coil to perform zero-return compensation on the horizontal component.

[0011] Furthermore, the control processor projects the zero-return compensation vector onto the two equivalent magnetic field directions represented by the coil magnetic field response fingerprint and performs projection decomposition. The projection decomposition is used to generate the first zero-return drive current and the second zero-return drive current. The control processor combines the first zero-return drive current and the second zero-return drive current into a zero-return drive current and drives the active magnetic compensation coil to form a zero-return compensation magnetic field in the zero-return sampling frame to obtain the zero-return geomagnetic vector. The control processor sequentially performs hard iron compensation and soft iron compensation on the zero-return geomagnetic vector to obtain the coil-compensated geomagnetic vector and writes it into the memory.

[0012] Furthermore, during the initialization phase, the control processor reads the sensor calibration data and installation extrinsic parameter data from the memory, performs zero-bias compensation, scale compensation, and non-orthogonal compensation on the angular velocity and acceleration sequences, and unifies them to the glasses' body coordinate system. It also performs hard iron compensation and soft iron compensation on the geomagnetic vector output by the geomagnetic sensor and unifies it to the glasses' body coordinate system. Based on the acceleration sequence, the control processor extracts the gravity direction and generates the initial pitch and initial roll. Based on the geomagnetic vector and gravity direction, it generates the initial heading and forms the initial head attitude value. The control processor establishes the mapping relationship between the homing target geomagnetic vector, coil geometric parameters, coil drive current, and coil magnetic field response, displays the coordinate extrinsic parameters, and establishes the navigation target retrieval rules.

[0013] Furthermore, the attitude fusion module performs attitude propagation based on the angular velocity sequence to obtain the propagated head attitude, and extracts the gravity direction based on the acceleration sequence to perform pitch and roll corrections on the propagated head attitude to form a corrected head attitude; the robust magnetic update module extracts the magnetic heading observation direction based on the coil-compensated geomagnetic vector and the gravity direction, and constructs the heading deviation amount with the heading direction corresponding to the corrected head attitude; the robust magnetic update module performs piecewise constraint correction on the heading deviation amount based on Huber anti-outlier loss, the piecewise constraint correction includes: converting the heading deviation amount into a correction amount candidate and according to the first segment interval stored in the memory and the second segment interval stored in the memory. The two-segment interval determines the interval to which the correction quantity candidate belongs. When the correction quantity candidate falls into the first segment interval, a smooth correction quantity is generated according to the smooth mapping table stored in the memory. When the correction quantity candidate falls into the second segment interval, a restricted correction quantity is generated according to the linear mapping table stored in the memory and the restricted correction quantity is limited to the heading correction quantity limit range stored in the memory. The robust magnetic update module uses the smooth correction quantity or the restricted correction quantity as the heading correction quantity to perform an equivalent rotation update around the gravity direction to obtain the head attitude. The equivalent rotation update includes converting the heading correction quantity into a quaternion increment and performing normalization composition with the quaternion representation of the corrected head attitude.

[0014] Furthermore, the direction recognition module generates the gaze direction based on the head posture and display coordinate extrinsic parameters; the position acquisition module outputs the current position; the content triggering module retrieves the set of guide targets from the guide content library based on the current position and converts it into a set of relative orientations; the content triggering module generates candidate guide targets based on the gaze direction and the set of relative orientations and performs a dwell determination to form a trigger confirmation state; when the trigger confirmation state meets the trigger conditions stored in the memory, it generates a guide content instruction and controls the display component to output the guide content.

[0015] By employing the above technical solution, this invention achieves the following beneficial effects: By setting active magnetic compensation coils around the geomagnetic sensor and driving the coils to operate according to a phase-encoded zero-return sequence by a control processor, the geomagnetic sensor can still obtain repeatable and interpretable coil-compensated geomagnetic vectors even under complex wearing environments and external magnetic disturbance conditions, thereby reducing the impact of magnetic disturbances on heading estimation from the source. Compared with methods that rely solely on static calibration or simple anomaly removal, this method ensures that geomagnetic observations always operate under controlled magnetic field conditions, and heading updates no longer frequently fail or jump due to environmental changes.

[0016] In terms of attitude calculation, this invention continuously performs attitude propagation through an inertial measurement unit and uses acceleration sequences to extract the gravity direction for pitch and roll correction, ensuring the continuity of head attitude during rapid head turns and walking undulations. Simultaneously, based on Huber's anti-outlier loss, it performs heading updates on the coil-compensated geomagnetic vector, achieving smooth correction when magnetic observation deviations are small, and constraining the correction amplitude when deviations are large. This effectively suppresses the excessive influence of a single abnormal magnetic observation on the heading results, thus achieving a balance between stability and responsiveness. This segmented constraint-based heading update method makes the direction recognition results more controllable in dynamic scenarios, reducing heading jitter caused by magnetic disturbances or transient noise.

[0017] At the guidance triggering level, this invention provides a continuous current position and constructs a structured guidance content library through a guidance resource component. It generates a gaze direction based on head posture and performs dwell determination based on geometric consistency, ensuring that guidance content triggering is based on the dual conditions of continuous alignment and spatial proximity. This approach avoids false triggering caused by momentary alignment, making the appearance of guidance content more consistent with the user's natural observation behavior. The display component outputs guidance content under the constraint of the gaze direction, providing the user with cues consistent with the actual direction within their field of vision, reducing the cost of understanding and improving the readability and credibility of the guidance information. Attached Figure Description

[0018] Figure 1 A schematic diagram of the timing control of the phase encoding return-to-zero sequence performed by the sensing compensation component provided in this embodiment of the invention and the corresponding geomagnetic sensor response waveform; Figure 2 This is a schematic diagram illustrating the vector projection decomposition principle for generating the zero-return drive current, provided in an embodiment of the present invention. Detailed Implementation

[0019] All features disclosed in this specification, or all steps in all disclosed methods or processes, may be combined in any way, except for mutually exclusive features and / or steps.

[0020] Any feature disclosed in this specification (including any appended claims and abstract) may be replaced by other equivalent or similar features, unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is merely one example of a series of equivalent or similar features.

[0021] A glasses-based navigation direction recognition and content triggering system integrating geomagnetism and inertial navigation includes: a sensing compensation component, a computing and storage component, a navigation resource component, and a display component. The sensing compensation component includes an inertial measurement unit, a geomagnetic sensor, an active magnetic compensation coil, and a coil drive module. The computing and storage component includes a control processor and a memory. The control processor drives the active magnetic compensation coil to obtain the coil-compensated geomagnetic vector according to the phase-encoded zero-return sequence, and performs attitude propagation based on the inertial measurement unit and performs heading update on the coil-compensated geomagnetic vector based on Huber anti-outlier loss to obtain the head attitude. The navigation resource component provides the current position and stores the navigation content library. The control processor generates the gaze direction based on the current position and head attitude, performs dwell determination to trigger the corresponding navigation content in the navigation content library, and outputs it through the display component.

[0022] The sensing compensation component within the glasses performs two main tasks: firstly, it continuously outputs angular velocity and acceleration sequences from the inertial measurement unit (IMU) to stably provide information on gravity direction and head rotation; secondly, it outputs a geomagnetic vector from the geomagnetic sensor within a controlled magnetic environment of the active magnetic compensation coil, ensuring the geomagnetic vector remains interpretable and repeatable in spatial direction, thus providing controllable input for subsequent heading updates. The advantage of placing the IMU, geomagnetic sensor, active magnetic compensation coil, and coil drive module within the same sensing compensation component is that the drive current establishment, geomagnetic sensor sampling window, and IMU timestamp can be aligned within the same time base, reducing transient magnetic disturbances and attitude propagation inconsistencies caused by asynchronous sampling.

[0023] The inertial measurement unit (IMU) preferably employs an integrated device combining a three-axis gyroscope and a three-axis accelerometer, mounted at a fixed position within the eyeglasses' coordinate system. To ensure the angular velocity sequence remains distortion-free during rapid head turns, the IMU's sampling frequency is preferably between 400 Hz and 1000 Hz, with one implementation using 800 Hz. Using the same sampling frequency for the acceleration sequence as the angular velocity sequence simplifies the timing matching for gravity direction extraction. The IMU output is sent to the control processor via a digital interface. The output data includes sampling time markers, which are used to establish a unified time reference with the synchronization trigger command from the geomagnetic sensor. The angular velocity sequence is used for attitude propagation because short-term heading changes are primarily caused by rotation around the gravity direction, and the angular velocity sequence provides continuous rotational increments. The acceleration sequence is used for gravity direction extraction because, during most periods of walking or stationary motion, the low-frequency components in the acceleration sequence approximate the gravity direction, suppressing the influence of gyroscope cumulative drift on pitch and roll, thus making the subsequent decomposition of the horizontal component of the geomagnetic vector more reliable.

[0024] The geomagnetic sensor is preferably a triaxial magnetic field sensor with a range covering the Earth's magnetic field and common environmental magnetic disturbances. In one embodiment, the geomagnetic sensor has a range of not less than 1000 microtesla and a sampling frequency of 100 Hz to 400 Hz, with one embodiment using 200 Hz. The geomagnetic sensor is placed in the eyeglasses body near the active magnetic compensation coil, ensuring that the compensation magnetic field generated by the active magnetic compensation coil has sufficient strength at the geomagnetic sensor. This allows the effective observation of the geomagnetic sensor to be pulled back to the vicinity of the zero-homing target geomagnetic vector even in the presence of external interference such as high-current devices, speaker magnets, and charging coils. The geomagnetic sensor is preferably placed away from speaker magnets and high-current loops, such as near the bridge or nose pad area, and the return loop area near the geomagnetic sensor is reduced at the printed circuit board wiring level to lower near-field magnetic disturbances caused by power supply ripple and radio frequency emissions.

[0025] An active magnetic compensation coil is positioned around the geomagnetic sensor. The active magnetic compensation coil includes a first compensation coil segment and a second compensation coil segment. The first and second compensation coil segments generate two linearly independent equivalent magnetic field directions at the geomagnetic sensor, thereby supporting the projection decomposition of the zero-homing compensation vector. In one embodiment, the first compensation coil segment employs a planar helical coil structure surrounding the geomagnetic sensor, with 40 to 120 turns, one embodiment having 80 turns; the second compensation coil segment employs a planar helical coil structure forming a geometric angle with the first compensation coil segment, with 40 to 120 turns, one embodiment having 80 turns. The two coil segments can be disposed on different layers of the same printed circuit board, utilizing interlayer alignment to achieve geometric stability. The consideration for using two coil segments is that: the compensation magnetic field generated by a single coil segment has a fixed direction and can only cancel magnetic disturbances in one direction, requiring multiple iterations when facing magnetic disturbances in any direction in space; the two coil segments provide two independent directions, which can directly obtain the first zero-return driving current and the second zero-return driving current through projection decomposition within one sampling period, thereby shortening the convergence time of the zero-return process and reducing the saturation risk of the geomagnetic sensor during the zero-return process.

[0026] The coil drive module is electrically connected to the first compensation coil segment and the second compensation coil segment. The coil drive module includes a constant current drive unit. The constant current drive unit outputs the drive current using a current closed-loop method. The closed-loop detection can be composed of a sampling resistor and an operational amplifier, or it can be composed of an integrated current detector and a digital-to-analog converter. The reason for using constant current output is that the magnetic field strength generated by the active magnetic compensation coil is approximately proportional to the coil current, while the coil resistance changes with temperature. If constant voltage drive is used, the current will drift with temperature, resulting in a non-repeatable coil magnetic field response fingerprint. Constant current drive separates the temperature-induced resistance change from the magnetic field strength, ensuring that the first and second excitation geomagnetic vectors still have stable changes under different ambient temperatures. In one embodiment, the drive current output by the coil drive module ranges from 0 mA to 40 mA. The first positive drive current is selected from a value between 10 mA and 25 mA, the second positive drive current is selected from a value between 10 mA and 25 mA, and the zero-return drive current is generated by the control processor based on the projection decomposition result and falls within the range of 0 mA to 40 mA. The coil drive module controls the drive current ripple under a current ripple constraint range, which is stored in memory. In one embodiment, the current ripple constraint range is limited to a peak-to-peak value of no more than 0.2 mA. The advantage of this is that the magnetic field response fingerprint is obtained by sampling from a geomagnetic sensor. The current ripple within the sampling window is regarded as an external magnetic disturbance by the geomagnetic sensor. The smaller the ripple, the closer the change between the zero magnetic vector and the excitation magnetic vector is to the true response of the coil, and the more stable the zero-return drive current obtained by projection decomposition is.

[0027] The phase-encoded zero-return sequence is organized into a zero-magnetic sampling frame, a first excitation sampling frame, a second excitation sampling frame, and a zero-return sampling frame within each sampling period. The zero-magnetic sampling frame corresponds to the coil drive module outputting a zero-drive current, and the geomagnetic sensor samples the zero-magnetic vector. The first excitation sampling frame corresponds to the coil drive module outputting a first positive drive current to the first compensation coil segment, and the geomagnetic sensor samples the first excitation geomagnetic vector. The second excitation sampling frame corresponds to the coil drive module outputting a second positive drive current to the second compensation coil segment, and the geomagnetic sensor samples the second excitation geomagnetic vector. The zero-return sampling frame corresponds to the coil drive module simultaneously outputting a zero-return drive current to both the first and second compensation coil segments, and the geomagnetic sensor samples the zero-return geomagnetic vector. Each sampling frame includes a drive current establishment phase and a geomagnetic sensor sampling window. The drive current establishment phase allows the constant current drive unit to enter steady-state output and attenuates the transient magnetic field and eddy currents near the coil. Placing the geomagnetic sensor sampling window after the drive current establishment phase reduces transient errors. In one embodiment, each sampling frame lasts for 5 milliseconds, with the drive current establishment phase lasting 2 milliseconds and the geomagnetic sensor sampling window lasting 3 milliseconds. The reason for placing the zero-magnetic sampling frame at the beginning of the sampling period is that the zero-magnetic geomagnetic vector serves as the reference within the same sampling period. This allows the current environmental magnetic field and the current structural magnetization state to be used as a common background, thus enabling the changes in the first excitation geomagnetic vector relative to the zero-magnetic vector, and the changes in the second excitation geomagnetic vector relative to the zero-magnetic vector, to more effectively reflect the direction and intensity of the coil's equivalent magnetic field, reducing the impact of slow environmental drift. The reason for placing the homing sampling frame at the end of the sampling period is that the homing drive current is derived from the zero-magnetic geomagnetic vector, the first excitation geomagnetic vector, and the second excitation geomagnetic vector. Placing it at the end ensures that the calculation of the homing compensation vector uses the coil's magnetic field response fingerprint within the same sampling period, avoiding additional errors introduced by cross-period drive drift.

[0028] refer to Figure 1 In a typical hardware implementation, Figure 1 The horizontal axis represents the time axis, measured in milliseconds. The vertical axes represent the amplitude of the drive current output by the coil drive module and the amplitude of the magnetic induction intensity sampled by the geomagnetic sensor, respectively. The entire sampling period is strictly defined as the time length. In this embodiment It is set to 20 milliseconds. This cycle length is not arbitrary, but rather chosen to balance the persistence of human vision with the bandwidth of geomagnetic sensors, ensuring that the rendering frame rate of subsequent guided content matches the pose update rate. For example... Figure 1 As shown, the control processor divides a complete sampling period into four consecutive logical time slots in the time domain: a zero-magnetic sampling frame, a first excitation sampling frame, a second excitation sampling frame, and a return-to-zero sampling frame. The duration of each time slot is equal and is set to... That is, 5 milliseconds. At the beginning of the timing sequence, during the zero-magnetic sampling frame, the control processor sends a command to the coil drive module to force the drive current. The value remains zero. At this point, the geomagnetic sensor is in an environment free from active magnetic field interference, primarily picking up the ambient background magnetic field and the remanent magnetism of the hard iron within the glasses. The sampled values ​​during this stage are recorded as the zero-magnetic geomagnetic vector. The purpose of setting this reference frame is to establish the magnetic environment background at the current moment so that common-mode interference can be eliminated in subsequent calculations through differential operations, especially eliminating the external environmental magnetic field that changes slowly as the wearer moves.

[0029] This is followed by the first excitation sampling frame and the second excitation sampling frame. During the first excitation sampling frame, the coil drive module injects a constant first positive drive current into the first compensation coil segment. The geomagnetic sensor samples the first excitation geomagnetic vector after the magnetic field stabilizes. Similarly, during the second excitation sampling frame, a second positive drive current is injected into the second compensation coil segment. The second excitation geomagnetic vector was obtained by sampling. . Figure 1 The micro-timing structure within each frame is clearly shown: the beginning of each frame is defined as the drive current setup phase. The duration is 2 milliseconds; the latter part is defined as the sampling window of the geomagnetic sensor. The duration is 3 milliseconds. (Introduction) The key lies in the physical considerations. Inductive devices have a current step response time, and the surrounding metal structures (such as frames and battery shields) generate induced eddy currents when the magnetic field changes abruptly. The reverse magnetic field generated by these eddy currents decays exponentially over time. If sampling occurs immediately at the moment of current switching, the geomagnetic sensor will capture unsteady data containing transient eddy current magnetic fields, leading to distortion in the subsequently calculated magnetic field response fingerprint. Therefore, this embodiment uses a hardware timer to force a wait before the geomagnetic sensor's sampling window opens. Exhausted, ensure and This reflects the steady-state magnetic field response. (After obtaining...) and Then, the control processor immediately performs vector operations internally to solve for the zero-return drive current for the current environment. Figure 1 At the end of the time axis, during the zero-return sampling frame, the coil drive module simultaneously outputs the calculated composite current to both sets of coils. The value measured by the geomagnetic sensor at this time is the zero-return geomagnetic vector. . Figure 1The magnetic field response curve in the image intuitively reflects this process: during the zero-return sampling frame, the magnetic induction intensity curves are not simply superimposed, but are adjusted to be close to the preset target zero-return value. This timing design strictly aligns the drive and sampling through synchronous trigger commands, ensuring the causality and temporal correlation of the data within the same period, and effectively suppressing attitude calculation errors caused by sampling jitter.

[0030] The synchronous trigger command is generated by the control processor within each sampling period and simultaneously sent to the coil drive module and the geomagnetic sensor. The synchronous trigger command constrains the coil drive module to switch between zero drive current, first positive drive current, second positive drive current, and homing drive current at a fixed phase point, while simultaneously constraining the geomagnetic sensor to complete sampling within a predetermined sampling window. The benefits of using the synchronous trigger command are twofold: Firstly, the geomagnetic sensor is more sensitive to transients in drive current switching. If the sampling window and the current establishment phase are misaligned, the transient will be treated as a change in the magnetic field, thus compromising the reproducibility of the coil's magnetic field response fingerprint. Secondly, the angular velocity sequence of the inertial measurement unit changes rapidly during rapid head rotation. The synchronous trigger command pins the geomagnetic sampling time to a specific position, allowing the control processor to use the angular velocity sequence near that time for attitude propagation interpolation, avoiding artificial increases in heading deviation caused by the geomagnetic observation direction and attitude propagation direction originating from different time slices.

[0031] The coil magnetic field response fingerprint is constructed within each sampling period from the zero-magnetic vector, the first excitation magnetic vector, and the second excitation magnetic vector. For ease of explanation, the changes and decomposition process are described using the following form, with a clear meaning given for each symbol. Let... Let the zero magnetic vector be denoted by . Let represent the first excitation geomagnetic vector, and let Let the second excitation geomagnetic vector be denoted as . Let represent the change in the first excitation geomagnetic vector relative to the zero geomagnetic vector. Let represent the change in the second excitation geomagnetic vector relative to the zero geomagnetic vector. pass get, pass Obtained. Symbol , , , , Each component represents a three-axis vector, with each component corresponding to the three-axis output of the geomagnetic sensor. The reason for using differential construction of the variation is that the environmental magnetic field and the static magnetization of the structure affect each other within the same sampling period. , , Having common superposition terms, the common terms can be canceled out by difference, and the change is closer to the direction and intensity of the equivalent magnetic field of the first compensation coil segment and the second compensation coil segment.

[0032] The zero-magnetic target vector is used to define the target direction that the geomagnetic sensor is expected to approach after compensation. To focus the zero-magnetic return process on the heading-related horizontal component, the control processor extracts the gravity direction based on the acceleration sequence output by the inertial measurement unit and decomposes the zero-magnetic vector into vertical and horizontal components. Let... The unit vector representing the direction of gravity, with the symbol... Let be a three-axis vector, its direction obtained by normalizing the low-frequency components of the acceleration sequence. Let The vector representation of the zero magnetic vector, i.e. .make To represent the vertical component, let To represent the horizontal component, then pass get, pass We obtain, where the symbols The result of the vector dot product is a scalar. The zero-magnetic vector is decomposed into... and The benefits are as follows: The heading is primarily determined by the horizontal component, while the vertical component contains magnetic tilt information and varies with geographical location. Directly applying the compensation target to all three axes would couple the homing process to magnetic tilt and wearing attitude disturbances, resulting in greater variations in the homing drive current. Concentrating the compensation on… This approach can maintain stable heading observations while reducing over-adjustment of the vertical component, improving the repeatability of the homing process during undulating terrain. The target geomagnetic vector is also decomposed into its horizontal and vertical components. The control processor generates a homing compensation vector based on the difference vector between the target horizontal component and the target horizontal component. Geometrically, the homing compensation vector points in the direction that pulls the current horizontal component toward the target horizontal component; therefore, the directional stability of the homing compensation vector directly determines the stability of the homing drive current.

[0033] Projective decomposition is used to derive the first and second zero-return drive currents from the zero-return compensation vector. Intuitively, the zero-return compensation vector is represented as... and A linear combination of . Let Let represent the vector representation of the zero-homing compensation vector in the three-axis space. Let the proportional coefficient corresponding to the first zero drive current be denoted as . The proportional coefficient corresponding to the second zero-drive current is indicated by the following: the goal is to make... Approaching in direction .symbol and As a scalar, it means scaling the unit excitation response of the first and second compensation coil segments to the amplitude required for zero-return compensation. An feasible solution method is to project the three-axis vector onto a two-dimensional plane orthogonal to the direction of gravity before solving, thus avoiding redundancy introduced by the vertical component. Specifically, for... , , Execute separately and Orthogonal projection yields , , ,in , , .symbol , , All are three-axis vectors, but all are located at the same level as... Within an orthogonal plane. Then, two non-collinear basis directions are chosen within this plane to represent it, for example, using... The direction of is the first basis direction, and the direction orthogonal to the first basis direction and still in the plane is the second basis direction, obtained by vector dot product. and The system of linear equations is then solved directly. A concrete implementation example is as follows: First calculate... express unit vector, ,in The vector norm is a scalar; then calculate... Indicates that they are located in the same plane and are To find orthogonal unit vectors, the approach is to first take... Then normalize to get , where the symbol This represents the cross product of vectors. Then... and Projected to coordinate system, to obtain scalar pairs and ,in , , , This forms a two-dimensional linear combination. and First, we obtain the answer from the second equation. Substituting this into the first equation, we get... .symbol , , , All are scalars, representing the components of each vector on a selected two-dimensional basis. This method compresses the three-axis problem into a two-dimensional plane related to the heading, avoiding the problems caused by... , The weak vertical component leads to numerical instability, while preserving the directional independence of the two coil segments. The control processor will... and The first zero-loop drive current and the second zero-loop drive current are mapped to the first zero-loop drive current and the second zero-loop drive current, and the first zero-loop drive current and the second zero-loop drive current are combined into the zero-loop drive current and output to the coil drive module, so that the zero-loop sampling frame obtains the zero-loop geomagnetic vector.

[0034] refer to Figure 2 This diagram is constructed within a specific two-dimensional coordinate system, defined as a horizontal plane strictly orthogonal to the direction vector of gravity. The physical basis for choosing this plane for calculations is that, for eyeglass navigation systems, the key factor affecting heading accuracy is the horizontal component of the geomagnetic vector, while the vertical component is easily affected by the wearer's head tilting movements. Therefore, Figure 2 All vectors in the diagram represent the projection of the three-dimensional magnetic field vector onto the horizontal plane. The origin of the coordinate system is marked as 0 (black dot). Physically, origin 0 represents the horizontal projection point of the zero-magnetic vector output by the geomagnetic sensor at the current moment after subtracting the vertical component. Using this origin 0 as a reference, the system transforms the complex magnetic field control problem into a geometric tracking problem in a plane.

[0035] Figure 2 The red solid arrow, originating from point O, represents the projection of the change in the first excitation geomagnetic vector relative to the zero geomagnetic vector onto the horizontal plane, denoted as . This vector characterizes the direction and intensity of the effective horizontal magnetic field generated by the first compensation coil segment on the geomagnetic sensor under unit test current excitation. Similarly, the blue solid arrow originating from the origin 0 represents the projection of the change in the second excitation geomagnetic vector relative to the zero magnetic vector onto the horizontal plane, denoted as [vector]. Due to the limitations of the internal space structure of the eyeglasses, the physical installation angles of the first compensation coil segment and the second compensation coil segment cannot usually be perfectly orthogonal. Therefore, the figure shows... and These are two vectors with an angle other than 90 degrees. Although they are not orthogonal, they are linearly independent and together constitute the physical basis describing this two-dimensional magnetic field space. The solid green arrow in the diagram, originating from point O, represents the target zero-return vector calculated by the system, denoted as... This vector connects the origin 0 to the target endpoint, and its geometric length and direction are determined by the difference between the current geomagnetic horizontal component and the preset zero-return target geomagnetic vector horizontal component. This visually indicates the total combined magnetic field that the active magnetic compensation coil needs to generate to "pull" the current magnetic field back to the target state. To accurately calculate the generated... The required drive current must be Decomposed into and In these two non-orthogonal directions, this linear combination process occurs... Figure 2 It is represented by a decomposition path that connects the beginning and end.

[0036] First, along the diagram The purple solid arrows extending in direction represent linear components. scalar This corresponds to the driving current proportionality coefficient of the first compensation coil segment. Next, starting from the end of the purple arrow, connect to the green arrow. The solid black arrows at the vertices represent linear components. scalar This corresponds to the driving current proportionality coefficient of the second compensation coil segment. The purple and black arrows together form the parallelogram law (or triangle law) path of vector addition, vividly illustrating the linear equation. The geometric meaning of . To efficiently solve the coefficients in the above equations in an embedded processor. and , Figure 2 A set of auxiliary coordinate systems is introduced, where is denoted as The black dashed line is an auxiliary axis, and its direction is intentionally set to coincide with the red vector. Completely overlapping; marked as The black dotted line is the same as Vertical orthogonal auxiliary axes. This construction method cleverly achieves decoupling: because exist The projection component in the direction is zero, and the complex two-variable simultaneous equations can be transformed into simple projection calculations. In the diagram, from the green vector... The vertex falls vertically towards The black dotted lines along the axis (labeled "orthogonal projection auxiliary lines") illustrate this projection solution process. The perpendicular line is... The intercept on the axis is marked as , representing the projection value of the target vector onto the first auxiliary basis vector. By combining exist Projection of direction and basis vectors exist , For the projection components of the direction, the control processor can avoid complex matrix inversion operations and quickly calculate the coefficients using only algebraic operations. and .

[0037] Before being written to memory, the homing magnetic vector undergoes hard iron compensation and soft iron compensation to obtain the coil-compensated magnetic vector. Hard iron compensation is used to counteract the fixed bias generated by the permanent magnetization material in the eyeglass body, while soft iron compensation is used to correct magnetic field distortion caused by surrounding magnetic materials. In one embodiment, hard iron compensation is achieved by subtracting the three-axis constant bias vectors, and soft iron compensation is achieved by three-axis linear transformation. The advantage of placing hard iron compensation and soft iron compensation after homing is that the active magnetic compensation coil first pulls the operating point back to the vicinity of the homing target magnetic vector, and the output of the magnetic sensor falls within the stable range. Linear compensation within this range is closer to its assumptions, reducing distortion amplification.

[0038] In one optional implementation, the order of the first excitation sampling frame and the second excitation sampling frame can alternate every few sampling cycles to average the difference in electromagnetic coupling directionality caused by drive switching. Another optional implementation splits the zero-magnetic sampling frame into a zero-magnetic sampling frame at the beginning of the sampling cycle and a zero-magnetic sampling frame at the end of the sampling cycle. A consistency check is performed using the two zero-magnetic samplings to form the zero-magnetic geomagnetic vector. When the difference between the two zero-magnetic geomagnetic vectors exceeds the zero-magnetic consistency threshold stored in memory, the control processor extends the drive current setup phase to allow for more sufficient attenuation of the transient magnetic field near the coil. Yet another optional implementation adds a verification sampling frame after the zero-return sampling frame. The coil drive module restores the zero drive current and samples the verification geomagnetic vector. By comparing the verification geomagnetic vector with the zero-return geomagnetic vector, it is determined whether the transient magnetization effect of the zero-return compensation magnetic field on the surrounding materials is negligible. If it is not negligible, the drive current setup phase of the zero-return sampling frame is extended and the drive switching speed is reduced, allowing the magnetization effect to attenuate before the sampling window.

[0039] In the above implementation, the sensing compensation component binds the driving process of the active magnetic compensation coil and the sampling process of the geomagnetic sensor to a unified time standard through a phase-encoded zero-return sequence. Then, it uses the coil magnetic field response fingerprint to characterize the equivalent magnetic field direction and intensity of the first and second compensation coil segments. Finally, it directly generates the first and second zero-return driving currents through projection decomposition, thereby obtaining the coil-compensated geomagnetic vector within a single sampling period. This processing reduces the sensitivity of the geomagnetic sensor output to environmental magnetic disturbances and to time misalignment caused by rapid head turns, and provides a stable and interpretable input for subsequent heading updates based on Huber anti-outlier loss.

[0040] The computational storage component consists of a control processor and a memory. The control processor is responsible for scheduling the time base, phase-encoded zero-return sequence, attitude propagation, heading updates, and head attitude output. The memory stores sensor calibration data, installation extrinsic data, current ripple constraint ranges, first segment intervals, second segment intervals, smoothing mapping tables, linear mapping tables, and heading correction limit ranges. It also caches angular velocity sequences, acceleration sequences, and coil-compensated geomagnetic vectors. The control processor and memory communicate via a bus. The control processor uses atomic writes or double buffering for critical memory areas to prevent the mixing of angular velocity sequences and coil-compensated geomagnetic vectors in different time slices, which could cause abnormal heading deviations.

[0041] In a feasible hardware configuration, the control processor uses a processor core with a clock speed in the range of 200 MHz to 800 MHz. The memory includes on-chip static random access memory (SRAM) and off-chip non-volatile memory (NVRAM). The on-chip SRAM is used for runtime caching, while the NVORAM stores sensor calibration data, installation extrinsic data, and mapping tables. To ensure the timing stability of the phase-encoded zero-return sequence, the control processor is configured with a hardware timer to generate fixed-period trigger pulses. These trigger pulses are used simultaneously to drive the coil drive module to switch the drive current and trigger the geomagnetic sensor sampling window. The direct benefits of this approach are: the relative positions of the drive current setup phase and the geomagnetic sensor sampling window are fixed within each sampling period; the changes in the coil's magnetic field response fingerprint are less likely to be contaminated by transients at the drive edge; and the repeatability of the coil's compensation for the geomagnetic vector to external magnetic disturbances is enhanced.

[0042] The phase-encoded zero-return sequence is implemented as a state machine within the control processor, driven by a hardware timer interrupt. In one implementation, the sampling period is set to 20 milliseconds. Within each sampling period, a zero-magnetic sampling frame, a first excitation sampling frame, a second excitation sampling frame, and a zero-return sampling frame are executed sequentially, with each sampling frame lasting 5 milliseconds. Each sampling frame is further divided into a drive current setup phase and a geomagnetic sensor sampling window. The drive current setup phase lasts 2 milliseconds, and the geomagnetic sensor sampling window lasts 3 milliseconds. Placing the drive current setup phase first allows the constant current drive unit to converge in a closed loop and attenuates eddy currents near the coil, preventing the geomagnetic sensor from measuring transient magnetic fields caused by current transitions within the sampling window. Placing the geomagnetic sensor sampling window later allows the integrated sampling of the geomagnetic sensor to more closely approximate the steady-state magnetic field, ensuring that the difference between the zero-magnetic geomagnetic vector, the first excitation geomagnetic vector, and the second excitation geomagnetic vector primarily reflects the coil response rather than edge transients.

[0043] The control processor outputs a zero-drive current to the coil drive module in the zero-magnetic sampling frame and triggers the geomagnetic sensor to sample and obtain the zero-magnetic vector. Subsequently, in the first excitation sampling frame, it outputs a first positive drive current to the first compensation coil segment and triggers the geomagnetic sensor to sample and obtain the first excitation geomagnetic vector. In the second excitation sampling frame, it outputs a second positive drive current to the second compensation coil segment and triggers the geomagnetic sensor to sample and obtain the second excitation geomagnetic vector. The selection of the first and second positive drive currents takes into account two factors: on the one hand, the change should be significantly higher than the noise of the geomagnetic sensor; on the other hand, it should avoid nonlinear magnetization of materials near the coil. In one embodiment, the first positive drive current is 18 mA and the second positive drive current is 18 mA. In the zero-return sampling frame, the control processor calculates the zero-return drive current and simultaneously outputs the zero-return drive current to the first and second compensation coil segments, triggering the geomagnetic sensor to sample and obtain the zero-return geomagnetic vector. After completing hard iron compensation and soft iron compensation, the zero-return geomagnetic vector is written into the memory as the coil compensation geomagnetic vector. Placing hard iron compensation and soft iron compensation after homing allows the compensation calculation to be completed within the undistorted operating range of the geomagnetic sensor. Linear compensation is closer to its applicable conditions, and coil compensation of the geomagnetic vector is more friendly to heading updates.

[0044] The coil's magnetic field response fingerprint is constructed from the zero-magnetic vector, the first excitation magnetic vector, and the second excitation magnetic vector. To make the differential calculation explicit and verifiable in software implementation, the control processor uses vector difference to generate the change. Let... Represents the zero magnetic vector. This represents the first excitation geomagnetic vector. This represents the second excitation geomagnetic vector. , , All are 3-dimensional vectors, with each dimension corresponding to one of the three axes of the geomagnetic sensor's output. Let... This represents the change in the first excitation geomagnetic vector relative to the zero geomagnetic vector. Let represent the change in the second excitation geomagnetic vector relative to the zero geomagnetic vector. Depend on get, Depend on The advantage of differential sampling is that, within the same sampling period, the environmental magnetic field and the static magnetization of the eyeglasses... , , Having common superposition terms, the common superposition terms are canceled out by difference. and This more comprehensively characterizes the direction and intensity of the equivalent magnetic field exerted by the first and second compensation coil segments on the geomagnetic sensor, thereby making the solution for the zero-return drive current more stable.

[0045] The zero-return compensation vector, constrained by the gravity direction within the control processor, acts only on the horizontal component. The gravity direction originates from the low-frequency portion of the acceleration sequence, and low-frequency extraction employs a discrete low-pass filter with fixed coefficients to avoid introducing historical data that requires long-term accumulation. In one implementation, the acceleration sequence sampling frequency is 800 Hz, and the equivalent time constant of the low-pass filter is set to 0.25 seconds. Let... This represents the acceleration vector after low-pass filtering. Let be a 3-dimensional vector; This represents the unit vector indicating the direction of gravity. If it is a 3-dimensional vector, then Depend on Received, among which The vector norm is a scalar used to represent vectors. Normalized to a unit vector. Used by the control processor. The zero-magnetic vector is decomposed into vertical and horizontal components. Let... The vector representation of the zero magnetic vector is taken as... ;make Indicates the vertical component. To represent the horizontal component, then Depend on get, Depend on Received, among which This indicates a vector dot product, with the result being a scalar. The benefit of applying zero-return compensation only to the horizontal component is that the heading corresponds more directly to the direction of the horizontal component. The vertical component carries information about the magnetic tilt angle and minor changes in wearing posture. If the vertical component is forcibly returned to zero, the zero-return drive current will change with head pitch and shaking, which will increase the noise of the coil compensating for the geomagnetic vector in the heading direction.

[0046] The control processor reads the zero-target geomagnetic vector back from memory and similarly decomposes it to obtain the target horizontal component. Let... This represents the geomagnetic vector of the homing target. Let be a 3-dimensional vector; Indicates the target level component. If we represent the vertical component of the target, then... and Press and , The same decomposition method is used to obtain it. The zero-return compensation vector is denoted as... , It is a 3-dimensional vector, defined as the difference of the horizontal component in the target direction: . The direction indicates the direction to pull the current horizontal component toward the target horizontal component. The magnitude of the zero-return compensation vector reflects the degree of deviation of the current horizontal component. Defining the zero-return compensation vector on the horizontal component ensures that the zero-return process mainly adjusts the heading-related components, reducing disturbances to non-heading components, and making the zero-return compensation magnetic field cleaner for heading updates.

[0047] The zero-return drive current is generated through projection decomposition. In software implementation, projection decomposition is solved using two-dimensional coordinates to ensure numerical stability and facilitate verification. The control processor first... , , Project it onto a plane orthogonal to the direction of gravity. Let... , , Let represent the projected vectors, then , , Then construct an orthogonal basis in the plane. Let express unit vector, ;make Indicates and Orthogonal vectors that lie in the plane ,in Represent the cross product of vectors; let express unit vector, Under this basis vector, and Decompose into scalar components: , , , .in , , , All are scalars. The proportionality constant is then solved. and , , This is a scalar used to scale the unit response of the two coil segments to the amplitude required for zero-return compensation. A specific solution order is: first calculate... , then calculate This solution method transforms the triaxial problem into a solution for two scalars in a plane, avoiding the ill-conditioned matrix caused by the weak vertical component. At the same time, it utilizes the two equivalent magnetic field directions provided by the first and second compensation coil segments, which can generate a stable zero-return driving current within one sampling period.

[0048] Control processor and The mapping is performed using a first zero-return drive current and a second zero-return drive current. The mapping employs a linear ratio and superimposed drive current boundary constraints. The upper and lower limits of the boundary constraints are determined by the capability of the coil drive module and are stored in memory for consistency checks. In one embodiment, both the first and second zero-return drive currents are limited to the range of 0 mA to 40 mA. The control processor then combines the first and second zero-return drive currents into a single zero-return drive current, simultaneously driving the first and second compensation coil segments during the zero-return sampling frame to form a zero-return compensation magnetic field, obtaining the zero-return geomagnetic vector and generating the coil compensation geomagnetic vector. To verify the effectiveness of the zero-return process, the control processor can calculate the horizontal component deviation of the coil compensation geomagnetic vector at the end of each sampling period and compare it with the target allowable range in memory. In one embodiment, the target allowable range is set so that the angle between the target horizontal component of the zero-return target geomagnetic vector and the horizontal component of the coil compensation geomagnetic vector does not exceed 5 degrees. If this angle is exceeded, the drive current setup phase of the next sampling period is extended to 3 milliseconds to improve the sufficiency of the zero-return compensation magnetic field entering a steady state.

[0049] Attitude propagation is performed cyclically at high frequency in the control processor, in parallel with the phase-encoded zero-return sequence from the geomagnetic sensor. In one implementation, the angular velocity sequence is sampled at 800 Hz, and the control processor updates the propagating head attitude every 1.25 milliseconds. The propagating head attitude is represented using quaternions, which are numerically stable under continuous rotation and do not produce Euler angle singularities. Let... Indicates the first The propagation head posture quaternion at the next update moment Let be a 4-dimensional vector; Indicates the first The angular velocity vector at the next update time. Let be a 3-dimensional vector, where each of its three dimensions corresponds to an angular velocity about one of the three axes in the coordinate system of the glasses body; let This represents the time interval between adjacent update times. It is a scalar quantity, with the unit being seconds; let This represents the angle scalar value within that time interval. ;make Represents the unit vector of the axis of rotation. .use and Constructing quaternion increments , A 4-dimensional vector: The square brackets indicate the concatenation of the scalar and vector parts of the quaternion. Propagation updates are implemented through quaternion composition: .symbol This represents quaternion composition operations. This represents a normalization operation used to maintain the quaternion length at 1, preventing the accumulation of numerical errors. This attitude propagation method closely follows the angular velocity sequence, providing continuous heading change estimates during rapid head turns and offering a predictive benchmark for subsequent heading updates.

[0050] Based on attitude propagation, the control processor extracts the gravity direction from the acceleration sequence and performs pitch and roll corrections on the propagating head attitude to form a corrected head attitude. The purpose of pitch and roll corrections is to align the gravity direction in the attitude representation, thereby making the extraction of the horizontal component of the geomagnetic sensor more accurate. In one embodiment, pitch and roll corrections construct a correction rotation using the angle between the target gravity direction and the estimated gravity direction, and apply the correction rotation to the propagating head attitude in an equivalent rotation update manner to finally obtain the corrected head attitude. No weights or historical data need to be introduced here; the correction rotation is directly determined by the low-frequency components of the current acceleration sequence and the propagating head attitude.

[0051] Heading updates are performed in the control processor on a geomagnetic sampling cycle. Each time a new coil-compensated geomagnetic vector is written to memory, the control processor calculates the magnetic heading observation direction and constructs the heading deviation. The magnetic heading observation direction is determined jointly by the coil-compensated geomagnetic vector and the gravity direction. Let... Indicates the geomagnetic vector compensated by the coil. Let be a 3-dimensional vector; This indicates the horizontal component of the geomagnetic vector compensated by the coil. If it is a 3-dimensional vector, then Then put This is converted to an angular expression representing the magnetic heading observation direction. In one implementation, the forward axis of the eyeglass body coordinate system is denoted as... The axis, the right axis is denoted as The axis, the upward axis is denoted as axis, express exist Components of the axis, express exist Components of the axis, , All are scalars. Magnetic heading observation direction angle. pass Received, among which This represents the arctangent function with quadrant information, and its output is in radians. The control processor simultaneously obtains the predicted heading angle from the corrected head attitude. The value is in radians. The deviation from the heading is denoted as... , In radians, Depend on get, This represents the angle wrapping function, constraining the difference to... Range, avoid crossing. False large deviations may occur. The heading deviation includes both the magnitude of the deviation and the direction of rotation, making it suitable for driving equivalent rotation updates.

[0052] The heading update based on Huber's anti-outlier loss is implemented within the control processor using segmented constraint correction. The segment intervals and mapping tables are stored in memory; during runtime, only table lookups and amplitude limiting are performed, without relying on historical data. The memory stores the first and second segment intervals. In one implementation, the first segment interval is 0 to 6 degrees, the second segment interval is 6 to 30 degrees, and the heading correction amplitude limit is 0 to 12 degrees. The control processor processes the heading deviation... Convert to correction quantity candidate , For degree, conversion through Completed, among which Pi (π) is a constant. Then compare... Boundaries with the first segment interval and the second segment interval: when When the signal falls within the first segment interval, the control processor reads the smoothing correction amount from the smoothing mapping table; when When the input falls into the second segment interval, the control processor reads the restricted correction amount from the linear mapping table and limits the restricted correction amount to within the heading correction amount limit. The smooth mapping table and the linear mapping table can be stored in the form of input interval endpoints and output correction amounts, and the control processor uses linear interpolation to complete the continuous output.

[0053] To facilitate practical implementation, a smooth mapping table is provided as an example: inputs are 0, 1, 2, 3, 4, 5, and 6 degrees, with outputs of 0, 0.9, 1.7, 2.4, 3.0, 3.5, and 3.9 degrees. This table outputs close to but slightly smaller than the inputs at small deviations, converges quickly within normal noise levels, and avoids fully converting transient fluctuations from the geomagnetic sensor into heading fluctuations. A linear mapping table is provided as an example: inputs are 6, 10, 15, 20, 25, and 30 degrees, with outputs of 3.9, 4.8, 6.0, 7.2, 8.4, and 9.6 degrees. This table maintains linear growth at large deviations, but with a growth slope less than 1, effectively limiting outlier geomagnetic observations to a controlled range and preventing heading updates from being excessively influenced by single anomalous observations. Huber's anti-outlier loss is manifested here as a smooth form within the interval and a restricted form outside the interval: it maintains a smooth correction with an approximate quadratic form in the small deviation region and a restricted correction with an approximate linear form in the large deviation region, thus balancing convergence speed and anti-outlier capability. The control processor implements this form by looking up a table, avoiding complex floating-point optimizations on embedded processors and facilitating reuse across different products.

[0054] The control processor uses the obtained smoothed or constrained correction as the heading correction and performs an equivalent rotation update around the gravity direction to obtain the head attitude. The equivalent rotation update is implemented using a quaternion incremental approach. Let... Indicates the heading correction amount. In radians, Obtained by converting degrees to radians. Let This represents the unit vector indicating the direction of gravity. Pick Construction around Quaternion increment , Then Perform normalization compounding with the head posture correction quaternion: .in This represents the quaternion used to correct head posture. This represents the updated head attitude quaternion. The advantage of applying correction around the gravity direction is that the heading update does not introduce additional pitch and roll disturbances, the gravity direction remains consistent, and the subsequent horizontal component decomposition and magnetic heading observation direction calculation maintain a stable closed loop, avoiding oscillations caused by distorting the gravity direction when correcting the heading.

[0055] In the above process, the memory does not handle historical learning; it only stores fixed constraints, tables, and a short-window cache. The short-window cache is used to solve the interpolation problem caused by inconsistent sampling frequencies. In one embodiment, the sampling frequency of the angular velocity sequence and acceleration sequence is 800 Hz, and the update frequency of the coil-compensated geomagnetic vector is 50 Hz to 200 Hz. When writing the coil-compensated geomagnetic vector, the control processor simultaneously records the timestamp of the sample and retrieves two or four sampling points near that timestamp from the angular velocity sequence cache for linear interpolation to obtain the angular velocity vector at the same timestamp, which is used for attitude propagation interpolation to predict the heading angle. Angle with magnetic heading observation direction With a consistent time base, the heading deviation reflects the true deviation rather than the timing deviation.

[0056] In an optional implementation, the control processor periodically alternates the frame order of the phase-encoded return-to-zero sequence. For example, in odd-numbered sampling periods, it executes a zero-magnetic sampling frame, a first excitation sampling frame, a second excitation sampling frame, and a return-to-zero sampling frame; in even-numbered sampling periods, it executes the same sequence. This averages the electromagnetic coupling directionality differences during drive switching over time, making the coil magnetic field response fingerprint closer to the stable response determined by the coil geometry. Another optional implementation adds a check sampling frame after the return-to-zero sampling frame. The coil drive module restores the zero-drive current and samples the check magnetic vector. The control processor compares the difference between the check magnetic vector and the zero magnetic vector to see if it falls within the zero-magnetic consistency threshold in memory. If not, the drive current setup phase of the next sampling period is extended to 3 milliseconds, and the first and second positive drive currents are reduced to 15 mA to reduce transient magnetization of surrounding materials. Another alternative implementation adjusts the boundary between the first segment interval and the second segment interval from 6 degrees to 4 degrees or 8 degrees to adapt to the magnetic disturbance intensity of different eyeglass structures. After adjustment, the smooth mapping table and the linear mapping table are updated synchronously, and the heading correction limit range remains unchanged or is synchronously adjusted to 10 to 15 degrees to ensure that the heading update remains controllable in a strong magnetic disturbance environment.

[0057] Through the above implementation, the control processor drives the active magnetic compensation coil to execute a phase-encoded zero-return sequence within a unified time base, forming a reproducible coil magnetic field response fingerprint and generating a coil-compensated geomagnetic vector. Simultaneously, attitude propagation is completed using an angular velocity sequence, pitch and roll corrections are completed by extracting the gravity direction using an acceleration sequence, and heading updates are completed using segmented constraint correction based on Huber anti-outlier loss, ultimately outputting the head attitude. The entire process can converge quickly in normal environments and limit the impact of single anomalies to the heading correction range when outlier magnetic disturbances occur, making it suitable for stable operation under the limited computing power and power consumption conditions of the glasses themselves.

[0058] The navigation resource component performs two tasks: providing the current location and storing the navigation content library, enabling the control processor to quickly retrieve the navigation target set based on the current location. To ensure the control processor can reliably trigger navigation content after generating a gaze direction based on the current location and head posture, the implementation of the navigation resource component typically prioritizes both the availability of location data and the deterministic nature of the navigation content library retrieval. The availability of location data determines the candidate range, while the deterministic nature of the navigation content library retrieval determines the continuity of the candidate set during movement. Continuity directly affects the stability of dwell time determination, ultimately impacting the repeatability of the trigger confirmation state.

[0059] The current location can be provided using multi-source positioning and uniformly represented within the navigation resource component. One implementation uses the latitude, longitude, and altitude output from the Global Navigation Satellite System (GNSS) receiver circuit as the current location, along with velocity and timestamp. To ensure smooth target set retrieval in glasses-based navigation scenarios, the navigation resource component sets the current location update frequency to 10 Hz to 30 Hz, with one implementation using 20 Hz. GNSS receiver circuits can produce jumps in obstructed environments, causing significant switching in the target set between adjacent updates, leading to frequent changes in candidate targets during dwell time. To reduce these jumps, the navigation resource component converts the current location to local planar coordinates before output and performs short-window smoothing. Short-window smoothing relies only on a fixed-length output buffer, with a buffer length fixed at 5 or 7 update cycles, and does not depend on long-term accumulated data. Smoothing employs median filtering or component speed limiting. One implementation performs median filtering on each component of the local planar coordinates, while another implementation sets a maximum allowable displacement of 5 meters for the displacement between two adjacent current positions to ensure that changes in the guide target set meet the interpretable range under pedestrian speed.

[0060] make Indicates the current location latitude. Indicates the current location's longitude. Indicates reference latitude, Indicates reference longitude. , , , All are scalars in radians. Let Represents the Earth's equivalent radius. This is a scalar quantity, with the unit being meters. Take 6378137. Local planar coordinates with eastward component. and northward component express, , All are scalars, with units in meters. A directly achievable approximation is... , The advantage of using local planar coordinates is that the distance calculation for the set of guided targets and the direction calculation for the set of relative orientations can be completed using addition, subtraction, multiplication, and division. The amount of computation is fixed and the values ​​are stable, making it easier to ensure the real-time performance of the control processor.

[0061] The navigation content library is stored in an itemized manner. Each item corresponds to a navigation target and includes a geographic location field, a trigger geometry field, and a content field. The geographic location field contains the latitude and longitude of the navigation target. The trigger geometry field contains the trigger radius, effective azimuth range, and selectable altitude range. The content field contains the rendering elements required by the navigation content instructions. The trigger radius is used to restrict the set of navigation targets to the vicinity of the current location. The effective azimuth range is used to constrain the navigation target to only enter the candidate navigation target list when it is near the gaze direction. One implementation sets the trigger radius to 30 meters to 120 meters, and another implementation uses 60 meters. The effective azimuth range is expressed using the center azimuth and the subtended angle. The center azimuth ranges from 0 degrees to 360 degrees, and the subtended angle ranges from 10 degrees to 60 degrees. One implementation uses a subtended angle of 24 degrees. The numerical trigger geometry field provides a consistent benchmark for the construction of trigger conditions, facilitating the formation of a stable trigger confirmation state during dwell determination.

[0062] The organization of the navigation content library affects retrieval speed. One implementation divides the local planar coordinates into a grid index and stores a mapping table of grid identifiers to the item list in memory. The grid side length can be 20 meters, 50 meters, or 100 meters; one implementation uses 50 meters. When the current location falls into a grid, the navigation resource component directly reads and merges the item lists of the nine adjacent grids to obtain the navigation target set. The reason for using a grid index is that positioning errors are commonly 3 to 10 meters outdoors. The grid index absorbs the error within a fixed retrieval window, making the navigation target set change more continuously as it moves, the identity of candidate navigation targets more stable, and the dwell time more easily accumulated. Another optional implementation sorts the items by latitude and longitude and uses a binary search to determine the range, then performs distance filtering on the items within the range. This is suitable for devices with a navigation content library size of less than 5,000 items.

[0063] After obtaining the current position, the control processor retrieves a set of navigation targets from the navigation content library and converts it into a relative orientation set. The relative orientation set contains at least the distance and azimuth angle of each navigation target relative to the current position. Let... Indicates the local planar coordinates of the current position. Indicates the local planar coordinates of the navigation target. , , , All are scalar quantities, with the unit being meters. Distance is denoted as... , This is a scalar quantity, with the unit being meters. The azimuth is denoted as... , It is a scalar in radians. ,in This is the arctangent function with quadrant information. Use... The advantages are: the azimuth angle changes continuously when the guide target is located in any quadrant, and there will be no jump when crossing the coordinate axis; the determination of candidate guide targets near the gaze direction is more stable.

[0064] The gaze direction is generated by the control processor by combining head pose and display coordinate extrinsic parameters. The head pose is expressed in the computational storage component as a quaternion or rotation matrix, and the display coordinate extrinsic parameters are used to align the optical exit direction of the display component to the eyeglass body coordinate system. One embodiment expresses the display coordinate extrinsic parameters as a fixed rotation matrix and stores it in memory. Let... This represents the rotation matrix from the glasses' body coordinate system to the navigation coordinate system. for The matrix is ​​derived from the head pose. Let... This represents the rotation matrix from the display coordinate system to the glasses' body coordinate system. for The matrix comes from the explicit coordinate extrinsic parameters. Let... This indicates the reference gaze vector in the display coordinate system. As a 3D vector, one implementation method is to... Take as This represents the forward axis of the display component's optical system. The vector of the gaze direction in the navigation coordinate system is denoted as... , It is a 3-dimensional vector. By explicitly multiplying the extrinsic parameters of the display coordinates, the user's gaze direction can be taken as the reference point, using the center direction as the reference rather than the geometric forward direction of the glasses. This makes the overlay position of the navigation content in the display components more consistent with the actual line of sight, resulting in a more consistent navigation experience.

[0065] Candidate navigation targets are generated using the geometric consistency of the gaze direction and the relative azimuth set. One implementation projects the gaze direction onto a horizontal plane and compares it with the azimuth angle. Let... This represents the projection vector of the gaze direction onto the horizontal plane. It is a 3-dimensional vector. ,in The unit vector representing the direction of gravity. Let be the dot product of the scalar. The horizontal angle representing the direction of gaze. It is a scalar in radians. ,in , They are respectively The scalar components are in the east and north directions. The control processor calculates the angular difference for each guide target. , It is a scalar in radians. , For the angle wrap function, the angle difference is limited to .when Less than half the angle corresponding to the effective azimuth range, and the distance If the radius is smaller than the trigger radius, the guided target enters the candidate guided target list. The half-angle value is related to the field of view of the display component. If the half-angle is too small, the user will need to align very precisely to trigger it; if the half-angle is too large, it is easy to introduce multiple candidate guided targets. One implementation sets the half-angle to 12 degrees and the trigger radius to 60 meters, which can balance the trigger frequency and accuracy when roaming through streets.

[0066] Dwell time determination is used to transform momentary alignment into a repeatable trigger confirmation state. One implementation maintains a dwell timer within the control processor for each candidate guide target. The dwell timer increments only as long as the same candidate guide target remains valid during continuous updates. When the dwell timer reaches a trigger condition, the system enters the trigger confirmation state. The trigger condition is stored in memory and uses explicitly defined values; in one implementation, the trigger condition includes a dwell time of not less than 0.8 seconds and an angle difference... During the duration of stay, the temperature remained below 12 degrees Celsius, and the distance was... The dwell time remains less than 60 meters. Incorporating both angular difference and distance into the dwell determination is based on the consideration that using only angular difference might trigger prematurely on distant targets, while using only distance might trigger content inconsistent with the gaze direction on lateral targets. Combining both restricts triggering to semantics related to looking towards and approaching the target. To avoid frequent entry and exit from candidate guide targets at boundaries, an alternative implementation incorporates angular difference hysteresis into the dwell determination: 12 degrees for entering a candidate guide target and 16 degrees for exiting. This absorbs slight head tremors within the hysteresis band, resulting in a more stable trigger confirmation state.

[0067] Once the confirmation state is triggered, the control processor reads the navigation content from the navigation content library and generates navigation content instructions. These instructions use a rendering description that the display component can directly execute, avoiding secondary parsing at the display component level. One implementation represents the navigation content instructions as an element list, containing text elements, icon elements, arrow elements, and optional audio prompt elements. Each element includes screen coordinates, size, color, level, and duration. Screen coordinates use display coordinates and are determined by the gaze direction; fixing the navigation content near the center of the field of vision reduces eye strain. Another implementation represents the navigation content instructions as a template identifier plus parameters. The template identifier points to a preset rendering template, and the parameters include the target name, distance value, directional indication, and prompt statement. The display component generates the final image according to the template, suitable for scenarios with a large navigation content library and uniform content format.

[0068] The display component is responsible for overlaying the navigation content onto the user's field of vision in a readable manner while maintaining low latency. One implementation employs a microdisplay and waveguide structure. The microdisplay has a refresh rate of 60 Hz, a resolution of 640 x 480, and a brightness range of 100 candela per square meter to 2000 candela per square meter. This brightness range is adapted for indoor and outdoor use, and the control processor can select the brightness level based on ambient light sensor input; one implementation provides six brightness levels. Setting the refresh rate to 60 Hz keeps the visual ghosting of the navigation content during head movements within an acceptable range, while keeping power consumption at a level manageable for the glasses. After receiving navigation content instructions, the display component uses a double-buffered rendering strategy: a rendering buffer is used to generate complete frames, and a display buffer is used for scanning output; buffer switching is completed at vertical synchronization. Double buffering avoids tearing caused by half-frame updates, resulting in better image stability, especially when arrows and text are updated simultaneously.

[0069] To ensure the navigation content aligns with the real-world orientation, the display component translates the gaze direction into screen coordinate offsets during rendering. One implementation defines the horizontal and vertical field of view angles within the display coordinate system. Let... Indicates the horizontal field of view. Indicates the vertical field of view. , All are scalars in radians. One implementation method is set... It is 30 degrees. It is 18 degrees. Let... Indicates the angular difference of the candidate guided tour targets. It is a scalar quantity in radians. The screen horizontal offset ratio is denoted as... , As a scalar, .when When the value is 0, the navigation content is located horizontally centered. When the value is 1 or -1, the navigation content is located at the horizontal boundary. The advantage of this mapping is that the angle difference is linearly related to the screen position, and users can intuitively understand which way to turn their heads to align with the target through visual feedback. The dwell time determination is easier to meet, and the confirmation state is triggered faster. A similar approach can be used in the vertical direction, using the pitch angle difference derived from the head posture to form the vertical offset ratio.

[0070] In an optional implementation, the display component supports two modes: transparent overlay and occlusion overlay. Transparent overlay displays the navigation content semi-transparently in the center of the field of view, suitable for road scenes; occlusion overlay draws a semi-transparent background behind the text, with a transparency of 0.3 to 0.7 (0.5 in one implementation), suitable for enhancing readability against complex backgrounds. Another optional implementation outputs preview navigation content after the confirmation state is triggered. The preview duration is 0.2 to 0.5 seconds (0.3 seconds in one implementation). The preview content only displays icons and directional arrows before the full text description unfolds. This preview design allows users to confirm directional consistency before reading the text, helping to reduce information interference after accidental touches.

[0071] Through the above implementation, the navigation resource component provides the current location and navigation content library with a clear data structure and defined retrieval path. The control processor converts the current location into a set of navigation targets and a set of relative orientations, then combines this with head posture to generate a gaze direction and performs a dwell time determination to form a trigger confirmation state. When the trigger confirmation state meets the trigger conditions, a navigation content instruction is generated, and the display component outputs the navigation content under a fixed refresh and double buffering mechanism. The key to the entire chain lies in the fact that the continuity of the current location, the stability of the candidate navigation target's identity, the consistency of the dwell time determination timing, and the low-latency output of the display component together form a closed loop. The relationship between the user's head turning action and the navigation content feedback is stable, and the navigation content triggering is more likely to achieve repeatable and verifiable implementation results.

[0072] While specific embodiments of the present invention have been described above, those skilled in the art should understand that these specific embodiments are merely illustrative. Those skilled in the art can omit, substitute, and modify the details of the above methods and systems in various ways without departing from the principles and essence of the present invention. For example, combining the above method steps to perform substantially the same function and achieve substantially the same result according to substantially the same method falls within the scope of the present invention. Therefore, the scope of the present invention is defined only by the appended claims.

Claims

1. A glasses navigation direction recognition and content triggering system integrating geomagnetism and inertial navigation, characterized in that, include: The system comprises a sensing compensation component, a computing and storage component, a navigation resource component, and a display component. The sensing compensation component includes an inertial measurement unit, a geomagnetic sensor, an active magnetic compensation coil, and a coil drive module. The computing and storage component includes a control processor and a memory. The control processor drives the active magnetic compensation coil to obtain the coil-compensated geomagnetic vector according to the phase-encoded zero-return sequence, and performs attitude propagation based on the inertial measurement unit and performs heading update on the coil-compensated geomagnetic vector based on Huber anti-outlier loss to obtain the head attitude. The navigation resource component provides the current position and stores the navigation content library. The control processor generates the gaze direction based on the current position and head attitude, performs dwell determination to trigger the corresponding navigation content in the navigation content library, and outputs it through the display component.

2. The system as described in claim 1, characterized in that, The active magnetic compensation coil includes a first compensation coil segment and a second compensation coil segment disposed on the eyeglass body. The coil driving module includes a constant current driving unit and is electrically connected to the first compensation coil segment and the second compensation coil segment respectively. The constant current driving unit is used to output a constant driving current during the duration of each sampling frame corresponding to the phase-encoded return-to-zero sequence, and to control the driving current ripple within the current ripple constraint range stored in the memory.

3. The system as described in claim 2, characterized in that, The control processor organizes the phase-encoded zero-return sequence into a series of zero magnetic sampling frames, first excitation sampling frames, second excitation sampling frames, and zero-return sampling frames in each sampling period. The zero-magnetic sampling frame corresponds to the coil drive module outputting zero drive current and triggering the geomagnetic sensor to sample and obtain the zero-magnetic geomagnetic vector. The first excitation sampling frame corresponds to the coil driving module outputting the first positive driving current to the first compensation coil segment and triggering the geomagnetic sensor to sample and obtain the first excitation geomagnetic vector; The second excitation sampling frame corresponds to the coil driving module outputting a second positive driving current to the second compensation coil segment and triggering the geomagnetic sensor to sample and obtain the second excitation geomagnetic vector; The zero-return sampling frame corresponds to the coil drive module simultaneously outputting zero-return drive current to the first compensation coil segment and the second compensation coil segment, triggering the geomagnetic sensor to sample and obtain the zero-return geomagnetic vector.

4. The system as described in claim 3, characterized in that, The control processor generates a synchronization trigger instruction in each sampling cycle and sends it to the coil drive module and the geomagnetic sensor simultaneously. The synchronization trigger instruction is used to constrain the coil drive module's drive current establishment phase and the geomagnetic sensor's sampling window to be aligned within the same time base, and to constrain the zero magnetic sampling frame, the first excitation sampling frame, the second excitation sampling frame, and the return-to-zero sampling frame to be completed in a fixed frame order.

5. The system as described in claim 3 or 4, characterized in that, The control processor constructs a coil magnetic field response fingerprint based on the zero magnetic field vector, the first excitation magnetic field vector, and the second excitation magnetic field vector. The coil magnetic field response fingerprint includes at least the change in the first excitation magnetic field vector relative to the zero magnetic field vector and the change in the second excitation magnetic field vector relative to the zero magnetic field vector. The change in the first excitation magnetic field vector relative to the zero magnetic field vector is used to characterize the direction and intensity of the equivalent magnetic field exerted by the first compensation coil segment on the geomagnetic sensor. The change in the second excitation magnetic field vector relative to the zero magnetic field vector is used to characterize the direction and intensity of the equivalent magnetic field exerted by the second compensation coil segment on the geomagnetic sensor.

6. The system as described in claim 5, characterized in that, The control processor extracts the gravity direction based on the acceleration sequence output by the inertial measurement unit, and decomposes the zero-magnetic geomagnetic vector into a vertical component parallel to the gravity direction and a horizontal component orthogonal to the gravity direction. At the same time, it decomposes the zero-return target geomagnetic vector into a target vertical component and a target horizontal component. The control processor generates a zero-return compensation vector based on the difference vector between the target horizontal component and the horizontal component. The zero-return compensation vector is used to constrain the active magnetic compensation coil to perform zero-return compensation on the horizontal component.

7. The system as described in claim 6, characterized in that, The control processor projects the zero-return compensation vector to the two equivalent magnetic field directions characterized by the coil magnetic field response fingerprint and performs projection decomposition. The projection decomposition is used to generate the first zero-return drive current and the second zero-return drive current. The control processor combines the first zero-return drive current and the second zero-return drive current into a zero-return drive current and drives the active magnetic compensation coil to form a zero-return compensation magnetic field in the zero-return sampling frame to obtain the zero-return geomagnetic vector. The control processor sequentially performs hard iron compensation and soft iron compensation on the zero-return geomagnetic vector to obtain the coil-compensated geomagnetic vector and writes it into the memory.

8. The system as described in claim 1, characterized in that, During the initialization phase, the control processor reads the sensor calibration data and installation extrinsic data from the memory. It performs zero-bias compensation, scale compensation, and non-orthogonal compensation on the angular velocity and acceleration sequences and unifies them to the glasses' body coordinate system. It performs hard iron compensation and soft iron compensation on the geomagnetic vector output by the geomagnetic sensor and unifies it to the glasses' body coordinate system. The control processor extracts the gravity direction based on the acceleration sequence and generates the initial pitch and initial roll. It generates the initial heading based on the geomagnetic vector and gravity direction and forms the initial head attitude value. The control processor establishes the mapping relationship between the geomagnetic vector of the homing target, the coil geometric parameters, the coil drive current and the coil magnetic field response, displays the coordinate extrinsic parameters, and sets the navigation target retrieval rules.

9. The system as described in claim 1, characterized in that, The attitude fusion module performs attitude propagation based on the angular velocity sequence to obtain the propagated head attitude, and extracts the gravity direction based on the acceleration sequence to perform pitch and roll corrections on the propagated head attitude to form a corrected head attitude. The robust magnetic update module extracts the magnetic heading observation direction based on the coil-compensated geomagnetic vector and the gravity direction, and constructs the heading deviation amount with the heading direction corresponding to the corrected head attitude. The robust magnetic update module performs piecewise constraint correction on the heading deviation amount based on Huber anti-outlier loss. The piecewise constraint correction includes: converting the heading deviation amount into a correction amount candidate and, based on the first segment interval and the second segment interval stored in the memory... The segment interval determines the interval to which the correction amount candidate belongs. When the correction amount candidate falls into the first segment interval, a smooth correction amount is generated according to the smooth mapping table stored in the memory. When the correction amount candidate falls into the second segment interval, a restricted correction amount is generated according to the linear mapping table stored in the memory and the restricted correction amount is limited to the heading correction amount limit range stored in the memory. The robust magnetic update module uses the smooth correction amount or the restricted correction amount as the heading correction amount to perform an equivalent rotation update around the gravity direction to obtain the head attitude. The equivalent rotation update includes converting the heading correction amount into a quaternion increment and performing normalization composition with the quaternion representation of the corrected head attitude.

10. The system as claimed in claim 1, characterized in that, The orientation recognition module generates the gaze direction based on head posture and display coordinate extrinsic parameters; the position acquisition module outputs the current position; the content triggering module retrieves the set of guide targets from the guide content library based on the current position and converts it into a set of relative orientations; the content triggering module generates candidate guide targets based on the gaze direction and the set of relative orientations and performs a dwell judgment to form a trigger confirmation state; when the trigger confirmation state meets the trigger conditions stored in the memory, it generates a guide content instruction and controls the display component to output the guide content.

Images