Search

CN-122015823-A - Space self-correcting system and method based on physical truth value and communication perception integration

CN122015823ACN 122015823 ACN122015823 ACN 122015823ACN-122015823-A

Abstract

The invention provides a space self-correcting system and a space self-correcting method based on integration of physical truth and communication perception, wherein the system part comprises a multi-source acquisition module for acquiring observation data of terminal equipment, a physical truth extraction module for extracting a multi-dimensional physical truth object, an absolute physical reference frame construction module for establishing an absolute physical reference frame, a physical consistency audit module for carrying out consistency audit on pose solutions output by visual/inertia/communication signals, a strong constraint fusion and closed-loop optimization module for carrying out joint modeling and optimization solving on visual residual errors and multi-dimensional physical truth residual errors, and a forced homing execution module for projecting the pose solutions back to the absolute physical reference frame and executing coordinate system self-correcting.

Inventors

  • CHEN HUICHONG

Assignees

  • 武汉华创全息数字科技有限公司

Dates

Publication Date
20260512
Application Date
20260202

Claims (10)

  1. 1. The spatial self-correcting system based on physical truth value and communication perception integration is characterized by comprising the following modules: The multi-source acquisition module is used for acquiring observation data of the terminal equipment, wherein the observation data at least comprises one or more of an IMU, geomagnetism and air pressure, and also comprises one or more of GNSS, visual semantics and ISAC signals; The physical truth value extraction module is used for extracting a multidimensional physical truth value object from the observed data, wherein the multidimensional physical truth value object at least comprises a gravity truth value, and optionally comprises a geomagnetic truth value, a barometric truth value, a GNSS truth value and a signal truth value; The absolute physical reference frame construction module is used for establishing an absolute physical reference frame based on the multidimensional physical truth value object, wherein a vertical axis is defined by gravity truth values, a heading standard is defined by geomagnetic truth values and magnetic declination, an altitude axis and level state are defined by barometric truth values, and a global position anchor point is defined by GNSS truth values; The physical consistency audit module is used for carrying out consistency audit on the pose solutions output by the vision/inertia/communication signals, and the consistency audit at least comprises pose consistency audit, course consistency audit, height consistency audit, semantic-physical consistency audit and motion consistency audit; the strong constraint fusion and closed loop optimization module is used for carrying out joint modeling and optimization solution on the visual residual error and the multidimensional physical truth value residual error, and carrying out gating or weight redistribution according to an audit result; And the forced homing execution module is used for projecting the pose solution back to the absolute physical reference frame and executing the coordinate system self-correction when the audit conflict meets the trigger condition.
  2. 2. The system for spatially self-correcting errors based on integration of physical truth and communication perception according to claim 1, wherein the physical consistency auditing module comprises a gravity gesture auditing unit for calculating an included angle between an upper direction vector of a visual gesture and a gravity truth direction, and triggering the forced homing execution module to perform projection righting on the gesture when the included angle exceeds a preset threshold value so as to eliminate Pitch/Roll drift of visual positioning; The physical consistency auditing module further comprises a course absolute auditing unit, and is used for comparing a course angle output by the visual odometer with an absolute north reference defined by geomagnetic truth values, performing cross-checking by combining GNSS course or speed priors, and triggering course forced homing when the course deviation exceeds a preset threshold value so as to eliminate course accumulated drift; The physical consistency auditing module further comprises a barometric pressure altitude auditing unit and a vertical level state machine, and is used for calculating altitude variation based on barometric pressure true values and identifying vertical mutation events, wherein the vertical mutation events trigger digital map level switching, semantic map reloading or pose altitude homing, and the vertical mutation events are determined by coupling of barometric pressure change rate gradients and a vertical motion model instead of only relying on absolute barometric pressure values; The physical consistency auditing module further comprises a communication signal consistency auditing unit, and is used for extracting an arrival angle AoA, a flight time ToF, a multipath fingerprint or Doppler frequency shift of an ISAC signal, and carrying out consistency mutual inspection on the ISAC signal and a pose, a speed or a direction under the absolute physical reference frame, and triggering gating or forced homing when conflict exists.
  3. 3. The spatially self-correcting system based on integration of physical truth and communication awareness according to claim 2, further comprising: The low-power consumption wake-up module is used for continuously collecting a gravity true value, a geomagnetic true value or an air pressure true value in a low-power consumption state, and waking up the camera or the semantic recognition module to enter a high-precision positioning mode when a sudden change event meeting a preset physical feature code is detected; When the pose of the semantic object does not meet the gravity plane constraint or the height level constraint, the system performs at least one process on the semantic object, namely projecting the pose of the semantic object to the gravity plane, performing semantic pose reset, refusing rendering or forcedly hiding the virtual object so as to prevent virtual content which does not meet the physical constraint from entering a display link; The motion consistency audit is used for consistency comparison of the estimated speed of vision/inertia and the GNSS speed or the Doppler deduction speed of communication signals, and a motion reliability score is generated based on the comparison result; the strong constraint fusion and closed loop optimization module performs rejection gating or continuous weight reduction on the visual observation residual errors according to the physical consistency audit result, so that optimization solution preferentially meets the physical truth value residual error item; When audit conflict occurs, the strong constraint fusion and closed loop optimization module promotes weight of residual items of gravity, geomagnetism, air pressure, GNSS or communication signals, reduces weight of visual residual, and enables pose solution to converge to an absolute physical reference frame.
  4. 4. The spatial self-rectification system based on integration of physical truth and communication perception according to claim 1, wherein the triggering conditions of the forced homing execution module include, but are not limited to, any one of attitude angle deviation exceeding a threshold, heading deviation exceeding a threshold, altitude deviation exceeding a threshold, semantic target conflicting with a gravity plane, and motion consistency score being lower than a threshold; the forced homing execution module rotationally projects the current gesture back to a vertical axis defined by a gravity true value so as to realize Pitch/Roll righting; The forced homing execution module projects the heading angle back to an absolute north reference or a GNSS heading reference defined by a geomagnetic true value and a magnetic declination to realize the elimination of the Yaw drift; and the forced homing execution module carries out homing on the pose height axis according to the height output calculated by the air pressure true value and drives the vertical level state machine to carry out map level switching.
  5. 5. The system for spatially self-correcting errors based on integration of physical truth and communication perception according to claim 3, wherein when the GNSS truth value meets the trusted condition, the forced homing execution module projects the position solution to the global anchor point neighborhood and reinitializes the local map to eliminate drift increment; When the communication signal consistency audit judges that the visual movement is not credible, the system uses geometric constraint derived by AoA/ToF or motion constraint derived by Doppler as homing reference to execute forced correction on pose solution; The strong constraint fusion and closed loop optimization module constructs the following joint optimization objective function: , And in any physical residual term When the visual residual error gating and the physical residual error weight lifting are performed and forced homing is triggered if necessary; The heterogeneous terminals are aligned on the basis of the gravity alignment vertical axis, the geomagnetic alignment Ji Hangxiang axis, the air pressure alignment hierarchical state and the GNSS alignment global anchor point of the same geographic area, and under the condition that the vision characteristic point cloud is not required to be shared, the absolute coordinate dynamic alignment of the whole life cycle is realized, and the dynamic alignment comprises initial alignment and asynchronous consensus calibration generated by drift in the running process; the communication signal consistency auditing unit is further used for generating an environment feature code based on the multipath fingerprint and assisting relocation, scene consistency auditing or fraud detection; The physical feature code of the low power wake-up module includes, but is not limited to, an abrupt air pressure change, an abrupt magnetic field fingerprint change, or an abrupt attitude inclination change, and a rapid physical consistency audit is performed after wake-up to decide to enter a high-precision positioning mode or return to a low power consumption state.
  6. 6. The spatial self-rectification system based on integration of physical truth and communication perception of claim 1, further comprising a spatial positioning rectification evidence obtaining system, wherein when the spatial positioning rectification evidence obtaining system detects that visual/inertial pose solution collides with a physical reference frame and triggers forced homing, rendering hierarchy or spatial anchor points of virtual content generate non-smooth fracture type jump, and an audit log corresponding to the jump is generated for external evidence obtaining; The audit log at least comprises a triggering reason category, a conflict residual value, a shielded observation item type and pose difference components before and after homing; The audit log includes, but is not limited to, a jump trigger time stamp, a residual error threshold judgment result, a gate control action zone bit and pose abstracts before and after homing, and can generate a tamper-proof evidence by adopting a local trusted execution environment or signature abstracting mode.
  7. 7. The spatially self-correcting system based on integration of physical truth and communication perception according to claim 1, further comprising an anti-interference spatially positioning audit system, the anti-interference spatially positioning audit system comprising: the motion consistency verification module is used for comparing the air pressure change rate with the IMU vertical displacement gradient and/or the visual odometer vertical displacement gradient in real time, judging that the air pressure observation is environment noise and executing nonlinear truncation when the consistency is lower than a threshold value, and shielding or substantially eliminating the correction authority of the air pressure factor to the height axis; When the magnetic field gradient exceeds a preset threshold value, judging that the magnetic field gradient is in a metal or electromagnetic interference environment, reducing geomagnetic constraint weight to approach zero, and simultaneously switching to a prediction mode based on visual relative heading and historical geomagnetic trend; the physical true value recovery mechanism is used for triggering the physical true value to reinitialize and executing once coordinate system correction after the environmental noise disappears so as to lead the positioning solution to be converged to an absolute physical reference frame again; the anti-interference space positioning audit system outputs interference level according to environmental fingerprints and dynamically adjusts confidence degree weights of physical truth value constraint items and communication perception constraint items so as to maintain self-correcting stability under the environment of abnormal magnetic fields or signal multipath; When the visual residual items and the physical residual items have directional conflict in the optimization updating direction, the system automatically shields or freezes the weight of the corresponding visual observation item through a physical audit logic gate, and promotes the weight of a physical truth value constraint item, so that the solving result is converged to an absolute physical reference frame; In the case of a terminal that does not include a camera or does not perform visual positioning, the system derives a displacement increment still based on the doppler shift, the phase change rate or the time of flight ToF of the communication signal, and carries out vertical axis and heading axis calibration and consistency auditing of the displacement increment under an absolute physical reference frame APR, triggering gating or forced homing when there is a conflict.
  8. 8. The spatial self-correcting system based on physical truth and communication perception integration of claim 1 is characterized by further comprising a spatial positioning auditing system without external base station dependence, wherein the spatial positioning auditing system without external base station dependence establishes an absolute geographic coordinate system by extracting a gravity vector and magnetic declination compensation, a double-threshold abnormal rejection mechanism is adopted by the system, a first threshold is based on the absolute deviation of sensor reading, a second threshold is based on the logical consistency of the physical quantity change rate and the inertia/visual displacement change rate, the system dynamically adjusts a weight matrix in a joint cost function according to the consistency, and the drift error of visual positioning is constrained by the aid of the non-drift characteristic of a physical constant so as to achieve error convergence under the whole scene.
  9. 9. The system for spatially self-correcting a signal based on integration of physical truth and communication perception according to claim 1, further comprising a spatial location convergence system based on a lie algebraic error state, the spatial location convergence system based on the lie algebraic error state comprising: The error state filtering and joint cost function construction module is used for taking pose output of the visual odometer as a prediction state, taking a gravity vector residual error, a geomagnetic vector residual error and an air pressure height residual error as physical observation constraint, and constructing a joint cost function comprising a visual re-projection residual error and a physical field vector residual error; The information matrix dynamic adjustment module is used for monitoring the spatial gradient and motion consistency of the sensor data in real time according to the environmental noise audit result and dynamically adjusting the information matrix weight of the physical field vector residual error in the joint cost function; And the forced convergence module is used for carrying out closed-loop forced constraint on the scale error and the heading error of the visual positioning by utilizing the non-drift characteristic of the physical constant, so that the covariance of the total error of the system is limited in a physical observation sphere and long-term stable convergence is maintained.
  10. 10. A spatially self-correcting method using the spatially self-correcting system according to claim 1, characterized in that the spatially self-correcting method comprises the steps of: S1, acquiring IMU, geomagnetism, air pressure, GNSS, visual semantics and communication perception ISAC signal observation data; s2, extracting a gravity true value, a geomagnetic true value, a barometric true value, a GNSS true value and a signal true value; s3, constructing an absolute physical reference frame, defining a vertical axis by a gravity truth value, defining a heading reference by a geomagnetic truth value, defining an altitude and level state by an air pressure truth value, and defining a global anchor point by a GNSS truth value; S4, performing physical consistency audit on visual/inertial/communication signal pose solutions, wherein the physical consistency audit comprises pose audit, course audit, height audit, semantic audit and motion audit; s5, constructing a multi-mode strong constraint optimization objective function and executing gating and weight redistribution; S6, when the audit conflict meets the triggering condition, executing forced homing to enable the pose solution to project back to the absolute physical reference frame and outputting the final world pose 。

Description

Space self-correcting system and method based on physical truth value and communication perception integration Technical Field The invention relates to the technical field of space calculation and high-precision space positioning, in particular to a space self-correcting system and a space self-correcting method based on physical truth value and Communication perception integration (INTEGRATED SENSING AND Communication, ISAC). Background With the rapid development of spatial computing (Spatial Computing), augmented Reality (AR) and Mixed Reality (MR) applications, terminal devices achieve stable, continuous and reusable spatial positioning and coordinate system alignment in complex real environments, and have become basic capabilities for immersive interaction, spatial navigation, multi-person collaboration and intelligent task execution. In the prior art, the mainstream scheme generally adopts Visual real-time positioning and map construction (Visual SLAM) or Visual inertial odometer (Visual-Inertial Odometry, VIO) as a core positioning means, and based on the core positioning means, external information such as an Inertial Measurement Unit (IMU) and a Global Navigation Satellite System (GNSS) is fused to improve positioning robustness and accuracy. However, the above-mentioned positioning system with vision as a core is essentially in the "probability estimation" paradigm, and its positioning result is highly dependent on environmental texture, illumination condition, occlusion dynamics and feature matching quality, so that it is difficult to avoid error accumulation and coordinate system drift in long-term real-world operation. Especially in the scenes of travel guidance, market multilayer navigation, vehicle-mounted high-speed movement, urban outdoor large space, multi-equipment cooperation and the like, the following structural defects and key pain points still exist in the prior art. In the prior art, the final-audit closed loop of 'absolute coordinate trusted location' is difficult to realize, the problems of drift accumulation, cross-equipment coordinate island, unstable vertical dimension, lack of physical-level trust root and the like generally exist in the visual/inertial/signal location technology, and meanwhile, although a future communication perception integrated route provides a new observation dimension, if the final-audit constraint of multidimensional physical true value is lacked, the absolute consistency and auditability of a location result are difficult to ensure. Disclosure of Invention Aiming at the defects existing in the prior art, the invention aims to provide a spatial self-correcting system and a spatial self-correcting method based on integration of physical truth and communication perception so as to solve the problems in the background art, and the invention constructs an Absolute physical reference frame (Absolute PHYSICAL REFERENCE, APR), executes physical consistency audit and realizes closed-loop forced homing under conflict conditions, thereby providing stable, reliable and difficult-to-bypass Absolute coordinate positioning capability for AR/MR, and having intelligent and cross-equipment cooperative scenes. In order to achieve the purpose, the invention is realized by the following technical scheme that the spatial self-correcting system and the method based on physical truth value and communication perception integration comprise the following modules: The multi-source acquisition module is used for acquiring observation data of the terminal equipment, wherein the observation data at least comprises one or more of an IMU, geomagnetism and air pressure, and also comprises one or more of GNSS, visual semantics and ISAC signals; the physical truth value extraction module is used for extracting a multidimensional physical truth value object from the observed data, wherein the multidimensional physical truth value object at least comprises a gravity truth value, and optionally comprises a geomagnetic truth value, a barometric truth value, a GNSS truth value and a signal truth value; The absolute physical reference frame construction module is used for establishing an absolute physical reference frame based on the multidimensional physical truth value object, wherein a vertical axis is defined by gravity truth values, a heading standard is defined by geomagnetic truth values and magnetic declination, an altitude axis and level state are defined by barometric truth values, and a global position anchor point is defined by GNSS truth values; The physical consistency audit module is used for carrying out consistency audit on the pose solutions output by the vision/inertia/communication signals, and the consistency audit at least comprises pose consistency audit, course consistency audit, height consistency audit, semantic-physical consistency audit and motion consistency audit; the strong constraint fusion and closed loop optimization module is used for carrying out joint model