Search

CN-122017805-A - Automatic calibration method and system suitable for mass production laser radar

CN122017805ACN 122017805 ACN122017805 ACN 122017805ACN-122017805-A

Abstract

The invention provides an automatic calibration method and an automatic calibration system suitable for mass production of laser radars, in particular to a quantum laser radar. The method comprises the steps of collecting an optical axis deviation angle and a mechanical positioning error of a radar to be calibrated, establishing an initial association characteristic set by using a support vector machine, extracting peak delay and amplitude fluctuation characteristics of an electronic response signal, calculating a photoelectric coupling coefficient through cluster analysis, generating a radar difference vector by combining an offset vector, resolving a calibration parameter based on a modular length of the radar difference vector, and driving an actuating mechanism to finely adjust an optical module or write firmware compensation to complete closed loop calibration. The invention realizes quantitative modeling and soft-hard collaborative compensation of multiple physical field coupling errors, improves the calibration precision to millimeter level, effectively solves the ranging deviation caused by temperature drift of the quantum laser radar, and takes the production efficiency and consistency into account.

Inventors

  • YU YANWU
  • LIU SHENGRONG

Assignees

  • 深圳光秒传感科技有限公司

Dates

Publication Date
20260512
Application Date
20251223

Claims (10)

  1. 1. An automated calibration method suitable for mass-produced lidar, the method comprising: acquiring optical axis deviation angle data and mechanical positioning error data of a laser radar to be calibrated, which are positioned on a production line, by a high-precision sensor; Performing multidimensional feature mapping and classification processing on the optical axis deviation angle data and the mechanical positioning error data by using a support vector machine classification algorithm, identifying an instance to be calibrated with potential coupling deviation, and establishing an initial association characteristic set containing an assembly gap size and an offset vector; collecting an electronic circuit response signal output by the to-be-calibrated example under a standard test environment, and extracting a signal peak delay characteristic and a response amplitude fluctuation characteristic from the electronic circuit response signal; performing cluster analysis on the signal peak delay characteristics and the response amplitude fluctuation characteristics, calculating coupling coefficients of interaction influence between a quantized optical system and an electronic circuit, and generating radar difference vectors representing radar comprehensive error states by combining the offset vectors; Judging whether the module length of the radar difference vector exceeds a preset tolerance threshold, if so, calculating a calibration parameter according to the coupling coefficient, wherein the calibration parameter comprises an assembly gap adjustment amount for an optical module or a signal compensation coefficient for an electronic circuit; And driving an actuating mechanism to perform physical displacement adjustment on the optical module according to the calibration parameters, or writing the signal compensation coefficient into a firmware storage area of the laser radar to be calibrated so as to complete closed-loop calibration.
  2. 2. The method of claim 1, wherein the performing multi-dimensional feature mapping and classification processing on the optical axis deviation angle data and mechanical positioning error data using a support vector machine classification algorithm comprises: Constructing the optical axis deviation angle data and the mechanical positioning error data into two-dimensional characteristic input vectors; Mapping the two-dimensional feature input vector to a high-dimensional feature space by adopting a radial basis function; and calculating the distance between the mapped two-dimensional characteristic input vector and a hyperplane of a preset support vector machine, and marking the laser radar to be calibrated as a normal example, a linear deviation example or a nonlinear coupling deviation example according to the distance.
  3. 3. The method of claim 1, wherein said extracting signal peak delay characteristics and response amplitude fluctuation characteristics from said electronic circuit response signal comprises: carrying out multi-layer decomposition on the acquired response signals of the electronic circuit by adopting wavelet transformation, and filtering background noise; locating the peak time point of the denoised electronic circuit response signal, calculating the time difference between the peak time point and the emission trigger signal, and determining the time difference as the signal peak delay characteristic; and performing fast Fourier transform on the denoised electronic circuit response signal, calculating the standard deviation of the amplitude distribution of the signal in the frequency domain, and determining the standard deviation of the amplitude distribution as the response amplitude fluctuation characteristic.
  4. 4. The method of claim 1, wherein calculating the coupling coefficient quantifying the interaction between the optical system and the electronic circuit comprises: normalizing the fit-up gap size, the signal peak delay characteristic and the response amplitude fluctuation characteristic; grouping the normalized data by adopting a K-means clustering algorithm to obtain feature clusters under different error modes; For each of the feature clusters, fitting an influence slope of an optical axis deviation angle change to the response amplitude fluctuation feature by linear regression analysis, and determining the influence slope as the coupling coefficient.
  5. 5. The method of claim 1, wherein driving an actuator to adjust the physical displacement of the optical module according to the calibration parameter comprises: converting the assembly gap adjustment amount into a driving voltage signal of a piezoceramic actuator; The piezoelectric ceramic actuator connected to the optical module is driven to perform micron-sized stepping movement along the optical axis direction; continuously acquiring the response signal of the electronic circuit in the moving process, and updating the module length of the radar difference vector in real time; And stopping driving and locking the current position of the optical module when the updated module length of the radar difference vector reaches a minimum value.
  6. 6. The method of claim 5, further comprising an active environmental disturbance testing step after locking the current position of the optical module, the active environmental disturbance testing step comprising: controlling the temperature of the test environment to generate step change in a preset time window, and applying thermal stress disturbance to the laser radar to be calibrated; continuously collecting point cloud coordinate data output by the laser radar to be calibrated in the temperature change process; Calculating a drift gradient of the point cloud coordinate data along with temperature change, if the drift gradient exceeds a preset thermal stability threshold, generating a secondary fine adjustment instruction according to the direction of the drift gradient, and driving the piezoelectric ceramic actuator again to correct the size of the assembly gap; Controlling a vibrating table of a test environment to apply mechanical vibration to the laser radar to be calibrated at a preset frequency; monitoring the change of the signal-to-noise ratio of the response signal of the electronic circuit in a mechanical vibration state; And if the signal-to-noise ratio is lower than a preset standard, adjusting the vibration frequency of the test environment to match the resonance characteristic of the laser radar to be calibrated, and recording the resonance frequency at the moment as calibration reference data.
  7. 7. The method of claim 6, further comprising a dynamic path planning step based on batch consistency: Counting the mechanical positioning error data of the laser radars to be calibrated in continuous preset quantity, and calculating consistency variances among batches; If the consistency variance is lower than a first threshold, judging that the current batch is a high consistency batch, and automatically selecting a simplified calibration path, wherein the simplified calibration path only performs electronic circuit parameter compensation; if the consistency variance is higher than a second threshold, judging that the current batch is an abnormal fluctuation batch, and automatically selecting a high-intensity calibration path, wherein the high-intensity calibration path comprises the physical displacement adjustment and the active environment disturbance test step of the whole process; when the abnormal fluctuation batch is judged, extracting a feature vector with the largest deviation in the batch; and feeding the characteristic vector back to the upper-level production link to generate a tolerance adjustment instruction aiming at front-end machining equipment.
  8. 8. The method of claim 1, wherein writing the signal compensation coefficients into a firmware memory area of the lidar to be calibrated comprises: Based on the calibrated residual error data, constructing a multidimensional electronic compensation lookup table, wherein the multidimensional electronic compensation lookup table establishes a distance correction value mapping relation under different working temperatures and detection distances; burning the multidimensional electronic compensation lookup table into a nonvolatile memory of the laser radar to be calibrated through a communication interface; And configuring a main control chip of the laser radar to be calibrated to read real-time temperature and measurement distance during operation, and correcting real-time data according to the multidimensional electronic compensation lookup table.
  9. 9. The method of claim 1, further comprising the step of creating a closed loop feedback model prior to resolving the calibration parameters from the calibration parameters: Collecting historical calibration data, wherein the historical calibration data comprises the initial association characteristic set before calibration and the radar performance improvement index after calibration; training a mapping model from the initial set of correlation properties to optimal calibration parameters using a neural network; and predicting the optimal assembly clearance adjustment amount of the current instance to be calibrated by using the mapping model after training, and taking the optimal assembly clearance adjustment amount as an initial calibration parameter.
  10. 10. An automated calibration system adapted for mass production of lidar, the system comprising: The multidimensional data acquisition unit is arranged at a production line station and is configured to acquire optical axis deviation angle data and mechanical positioning error data of the laser radar to be calibrated, and acquire an electronic circuit response signal output by the laser radar to be calibrated under a standard test environment; A central processing computing center in communication with the multi-dimensional data acquisition unit, the central processing computing center programmed to operate the following logic modules: the classification mapping module is used for running a support vector machine algorithm, mapping the optical axis deviation angle data and the mechanical positioning error data to a preset deviation classification space and generating an initial association characteristic set containing an assembly clearance size and an offset vector; The coupling analysis module is used for extracting signal peak delay characteristics and response amplitude fluctuation characteristics from the electronic circuit response signals, calculating and quantifying coupling coefficients of the interaction influence of the optical system and the electronic circuit based on cluster analysis, and further constructing radar difference vectors; the decision generation module is used for calculating calibration parameters when the modular length of the radar difference vector exceeds a tolerance threshold value, wherein the calibration parameters comprise a physical assembly clearance adjustment instruction or a firmware signal compensation coefficient; An execution subsystem, controlled by the central processing computing center, comprising: The precision mechanical adjusting mechanism is used for responding to the physical assembly clearance adjusting instruction and executing displacement operation on the optical module of the laser radar to be calibrated; And the firmware burning interface is used for responding to the firmware signal compensation coefficient and executing data writing operation on the internal memory of the laser radar to be calibrated.

Description

Automatic calibration method and system suitable for mass production laser radar Technical Field The invention relates to the technical field of radio direction finding and navigation, in particular to an automatic calibration method and system suitable for mass production of a laser radar. The invention is particularly suitable for the production line calibration of Quantum LiDAR (Quantum LiDAR), solid-state LiDAR and mechanical LiDAR with extremely high requirements on assembly accuracy and signal-to-noise ratio. Background With the rapid development of automatic driving and intelligent sensing technologies, a laser radar (LiDAR) is used as a core sensor, and consistency and precision control in the mass production process of the laser radar are key bottlenecks for restricting the development of industries. In particular, for a quantum laser radar employing a single photon detection technology (such as a single photon avalanche diode SPAD or a silicon photomultiplier SiPM), the quantum laser radar has a high sensitivity characteristic of a single photon level, which makes the system meet the unprecedented high requirements on the stability of micro mechanical deformation and circuit response of an optical component. However, in the current mass production process, the multiple physical field coupling effect between the optical system and the electronic circuit makes it difficult for the conventional calibration method to satisfy the high precision requirement. In the prior art, a step-by-step calibration strategy of 'first dimming and then power-on' is generally adopted, and the mechanical position of an optical component and the signal parameter of an electronic system are respectively and independently adjusted, so that the dynamic influence of the optical-mechanical structure deviation on a photoelectric detection signal is not considered. Due to the small optical axis deflection angle and mechanical positioning error existing in the assembly process, the falling point deviation of the echo light spot on the photoelectric detector can be caused. For quantum lidar, this shift can cause distortion in the statistical distribution of photon arrival times, which in turn changes the rising edge characteristics of the signal, causing ranging delay (Walk Error) and amplitude fluctuations. Such cross-domain coupling effects make system level optimization impossible by means of only a single field of compensation, and small physical deviations tend to be amplified to significant false alarm noise or measurement drift. In addition, the calibration result in the static standard environment can not reflect the performance drift of the product in practical application caused by temperature change or mechanical vibration, so that the problem of measurement misalignment of the quantum laser radar under the complex working condition is caused. The current industry lacks a unified model capable of quantifying the association relation between the optical mechanical error and the electronic response, and has no effective mechanism for realizing the closed-loop correction of the software and hardware cooperation in a mass production environment. Therefore, it is needed to provide a calibration method capable of being automatically executed, giving consideration to the interaction of multiple physical fields and supporting real-time feedback adjustment, so as to improve the measurement accuracy and batch consistency of the laser radar, especially the quantum laser radar product. It should be noted that the information disclosed in the foregoing background section is only for enhancement of understanding of the background of the invention and thus may include information that does not form the prior art that is already known to those of ordinary skill in the art. Disclosure of Invention In view of the above, the invention provides an automatic calibration method and system suitable for mass production of a laser radar, which aims to solve the problems of insufficient calibration precision, difficult prediction of temperature drift and poor consistency of batch-to-batch performance caused by neglecting multi-physical field coupling effect between an optical system and an electronic circuit in the prior art, and realizes quantitative analysis and closed-loop correction of photo-mechanical-electric coupling errors by combining support vector machine classification, clustered regression modeling and soft-hard collaborative compensation mechanisms through constructing a full-automatic calibration flow of data acquisition-characteristic coupling analysis-closed-loop execution, thereby remarkably improving the ranging precision and long-term stability of the laser radar under complex working conditions and meeting strict requirements of mass production on efficiency and consistency. The embodiment of the invention provides an automatic calibration method suitable for mass production of a laser radar, which comprises the following s