A waveform characteristic parameter calculation method of a full-waveform laser radar

Abstract

This invention discloses a method for calculating waveform characteristic parameters of a full-waveform lidar, applicable to full-waveform lidar employing single-pulse time-of-flight ranging. The method starts calculation from the peak value of discretized waveform data containing echo pulses, employing an on-chip implementation-optimized scheme to calculate the characteristic parameters of the waveform arrival time. To simplify the calculation steps, the initial time is set to 0, and the actual initial time is added to the final result. After obtaining the result of this iteration, waveform data from two adjacent times are added to the left and right for the next iteration. Iteration stops when the value of the newly added waveform data is less than 2 / 3 of the peak value and the difference between the arrival time result and the previous iteration is less than a preset minimum precision, or when the value of the newly added waveform data is less than a preset threshold. This method can obtain highly accurate measurement results and has the advantages of easy on-chip implementation and fast real-time computation.

Description

Technical Field

[0001] This invention belongs to the field of laser detection technology, specifically relating to a method for calculating waveform characteristic parameters of a full-waveform lidar. Background Technology

[0002] LiDAR, a combination of traditional radar technology and modern laser technology, utilizes the characteristics of lasers to improve upon the shortcomings of traditional radar technology, such as insufficient spatiotemporal resolution and limited detection range. It is widely used in land surveying, military reconnaissance, and industrial measurement. Common lidar ranging methods can be divided into triangulation ranging, interferometric ranging, and time-of-flight (TOF) optical ranging. Among these, pulsed TOF optical ranging has become the mainstream lidar ranging method due to its advantages of long detection range, high efficiency, and good ranging accuracy.

[0003] LiDAR systems employing pulsed optical time-of-flight ranging typically use simple analog methods to process echoes, such as fixed threshold methods and peak detection methods. Full-wave analysis is a novel waveform processing method that offers unparalleled resolution, sensitivity, and accuracy compared to analog methods. However, the full-wave analysis methods used in existing LiDAR systems can only be implemented offline, relying on high-performance computers and cannot be implemented on-chip. The computation process is complex and time-consuming, lacking advantages in portability and real-time performance.

[0004] Based on the above background, the present invention provides a method for calculating waveform characteristic parameters of a full-waveform lidar, which can realize real-time full-wave analysis on the chip through simple data processing, and achieve fast real-time calculation while ensuring accuracy. Summary of the Invention

[0005] The purpose of this invention is to provide a method for calculating the waveform characteristic parameters of a full-waveform lidar.

[0006] The technical solution to achieve the objective of this invention is as follows: Firstly, this invention provides a method for calculating waveform characteristic parameters of a full-waveform lidar, comprising the following steps:

[0007] Step 1: Set a preset threshold z = η + λ·σ based on the statistical characteristics of the background noise in the echo, where η is the mean of the background noise, σ is the variance of the background noise, and λ is the variance coefficient.

[0008] Step 2: The lidar emits a pulse signal, and the pulse emission time is recorded as time 0. The receiver digitizes the received echo signal to obtain the original waveform data containing the pulse echo, and determines the maximum value z in the waveform data. max , will z max Extract all data points between the maximum and minimum values ​​at the corresponding time points, and denote them as z.j j = 1, 2, 3, ..., let n be the number of data points extracted, and let t0 be the minimum value at the time corresponding to all the maximum values;

[0009] Step 3: Calculate the following 6 coefficients based on the number of data points n:

[0010]

[0011] Substitute the coefficients and data values ​​into the following calculation formula:

[0012]

[0013] The pulse arrival time μ calculated in this operation is obtained;

[0014] Step 4: Add an original waveform data point, denoted as z, to each side of the data used in the previous iteration. j j = 1, 2, 3, ..., let n be the number of data obtained, let t0 be the minimum value of the data at the corresponding time, repeat step 3 to get the new calculation result μ′ and calculate the difference |μ′-μ| between the result and the previous iteration result;

[0015] Step 5, if both new data added in step 4 are less than 2*z max If the value of the two new data added in step 4 is less than the preset threshold z, the iteration ends; otherwise, step 4 is repeated.

[0016] Step 6: Obtain the final feature parameter μ.

[0017] Furthermore, in step 1, before the lidar emits a pulse signal, the statistical distribution characteristics of the noise floor data obtained through the analog-to-digital converter are obtained, and a preset threshold is set based on the statistical distribution characteristics.

[0018] Furthermore, in step 2, the calculation starts from the maximum value of the echo data, and all data between the minimum and maximum values ​​within the time when the maximum value occurs are used as the data for the first iteration.

[0019] Furthermore, the formula for calculating the coefficient in step 3 is as follows:

[0020]

[0021] Based on the characteristics of pulsed lidar echoes, the minimum value in the data involved in the calculation does not affect the relative time difference between the final result and the value. Therefore, setting t0 = 0, the coefficient calculation formula simplifies to:

[0022]

[0023] Finally, adding t0 to the calculation result yields the correct feature parameters.

[0024] In a second aspect, the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method described in the first aspect.

[0025] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in the first aspect.

[0026] Fourthly, the present invention provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the method described in the first aspect.

[0027] Compared with the prior art, the significant advantages of the present invention are: addressing the problems of existing full-wave analysis methods relying on high-performance computers, having complex calculation processes and long computation times, and poor portability and real-time performance, the present invention provides a method for calculating waveform feature parameters of full-waveform lidar, which retains the advantages of traditional lidar in real-time convenience, while also possessing the high resolution, high sensitivity, and high precision characteristics of the full-wave analysis method. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of the calculation method of the present invention.

[0029] Figure 2 This is a schematic diagram of the hardware implementation of the present invention. Detailed Implementation

[0030] like Figure 1 As shown, this invention provides a method for calculating waveform characteristic parameters of a full-waveform lidar, comprising the following steps:

[0031] Step 1: Set a preset threshold z = η + λ·σ based on the statistical characteristics of the background noise in the echo, where η is the mean of the background noise, σ is the variance of the background noise, and λ is the variance coefficient, which is set according to the required accuracy and computational efficiency.

[0032] Step 2: The lidar emits a pulse signal, and the pulse emission time is recorded as time 0. The receiver digitizes the received echo signal to obtain the original waveform data containing the pulse echo, and determines the maximum value z in the waveform data. max , will z max Extract all data points between the maximum and minimum values ​​at the corresponding time points, and denote them as z. j Let j = 1, 2, 3, ..., and denote the number of data points extracted as n, and denote the minimum value at the time corresponding to all the maximum values ​​as t0.

[0033] Step 3: Calculate the following 6 coefficients based on the number of data points n:

[0034]

[0035] Substitute the coefficients and data values ​​into the following calculation formula:

[0036]

[0037] The result μ of this calculation is obtained.

[0038] Step 4: Add an original waveform data point, denoted as z, to each side of the data used in the previous iteration. j Let j = 1, 2, 3, ..., and denote the number of data points as n. Let the minimum value of the data points at the corresponding time be t0. Repeat step 3 to obtain a new calculation result μ′ and calculate the difference |μ′-μ| between the result and the previous iteration result.

[0039] In this step, based on the left-right symmetry of the pulse echo model, a new set of iterative data is added to each side of the previous iteration to form new iterative data. The new iterative result is closer to the actual value than the previous result.

[0040] Step 5, if both new data added in step 4 are less than 2*z max If the value of the two new data added in step 4 is less than the preset threshold z, the iteration ends; otherwise, step 4 is repeated.

[0041] Step 6: Obtain the final feature parameter μ.

[0042] This invention uses the function shown in the following equation to describe the time-domain image of the pulse echo:

[0043]

[0044] In the formula, δ is the time constant, and μ is the pulse arrival time. For lidar, μ is a key characteristic parameter for obtaining the measured distance. Taking the logarithm of the time-domain expression of the pulse echo and performing a Taylor expansion over the time variable t, the result can be written in the following form:

[0045] lny = b0 + b1t + b2t 2

[0046] The relationship between the polynomial coefficients and the characteristic parameters of the pulse echo is as follows:

[0047]

[0048] The Taylor expansion can be written in discretized matrix form, where the value of lny is replaced by z:

[0049]

[0050] Notation:

[0051] Z = XB

[0052] The value of the coefficient matrix B is obtained using the least squares principle:

[0053] (X T X)B=X T Z

[0054] B = (X) T X) -1 X T Z

[0055] The waveform characteristic parameter μ required for pulsed lidar can be calculated from the coefficients in the B matrix:

[0056]

[0057] As can be seen from the above full-wave analysis method, the process involves matrix multiplication and inversion, and the size of the matrix changes with the increase of the data volume n. Observing the above process, we know that matrix X is an n×3 matrix, therefore A = X. T X is a 3×3 matrix, as shown in the following formula:

[0058]

[0059] Its inverse matrix can be represented as A * / |A|. However, ultimately only the ratio of the last two data points in matrix B needs to be calculated, and |A| is eliminated, so the value of |A| can be ignored and not calculated. ∑t n k Write as a nk The process can be simplified to:

[0060]

[0061] Expanding further, the expression for the characteristic parameter μ can be obtained as follows:

[0062]

[0063] The above equation shows that no complex matrix operations are needed; only a finite number of polynomial calculations are required to obtain the result. However, as n increases, the number of required computational steps rises sharply. Therefore, to further optimize the algorithm, the above equation is further optimized. Based on the sampling principle of the ADC, we know that {t... n In fact, it is an arithmetic sequence with unknown initial values, that is:

[0064] t n =t fre (t0+k)k=1,2,3…,n-1

[0065] In the formula t fre As the sampling interval, a can be... n1 a n2 a n3 and a n4 Written as:

[0066]

[0067] The formula for summing natural numbers multiple times is:

[0068]

[0069] Substituting these values ​​into the original calculation formula, we can obtain a through only a finite number of calculations. n1 a n2 a n3 and a n4 The value of is further simplified to the following formula:

[0070]

[0071] The six fixed coefficients are:

[0072]

[0073] Based on the characteristics of pulse echoes, the value of the minimum moment in the data used for calculation does not affect the relative time difference between the final fitting result and the time difference. Therefore, we can set t0 = 0, and the coefficient calculation formula can be simplified to:

[0074]

[0075] Finally, adding t0 to the calculation result yields the correct result, significantly simplifying the calculation process. Compared to the original full-wave analysis method, the optimized waveform characteristic parameter calculation method greatly reduces computational complexity, lowers chip resource consumption and computation time, while maintaining the accuracy of the calculation results. Furthermore, the computational complexity does not change dramatically with the increase of data volume.

[0076] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0077] Example

[0078] This invention is applicable to digital full-waveform single-pulse lidar. In this embodiment, the wavelength of the pulsed laser emitted by the laser emitting device is 905 × 10⁻⁶. -9m, pulse width is 25×10 -9 s, peak power 35W, optical emission system aperture 0.045m, divergence angle 5×10 -3 The photoelectric receiving device is an APD detector with an A / W response of 0.65, an avalanche gain of 100, and an optical system receiving aperture of 0.040m. The high-speed echo data acquisition module uses a high-speed ADC as the acquisition core, discretizes the echo data at a sampling frequency of 1GHz, and obtains the 5GHz frequency data using CIC interpolation filtering in the FPGA.

[0079] The laser emitting device is turned off, and the photoelectric receiving device and the high-speed echo data acquisition module are turned on. After the FPGA receives the noise signal, it calculates the statistical characteristics of the noise data. The mean of the noise data is 0mV, the variance is 0.432mV, the variance coefficient is set to 3, and the required preset threshold is 1.296mV.

[0080] Then, the laser emitting device is turned on to emit a pulsed laser towards the target. After being reflected by the target at a standard distance of 50 meters, it is received by the photoelectric receiving device and converted into an electrical signal. This signal is then discretized by a high-speed ADC and processed through echo discrimination, interpolation, and filtering to obtain the raw waveform data, after which calculations begin. The calculation method is as follows: Figure 1 As shown, the first iteration starts with the maximum value in the waveform data. In this example, there are four time points where the waveform data has a maximum value of 270mV, with the minimum value occurring at t0 = 3.324 × 10⁻⁶. -7 There are 6 data points between the minimum and maximum time points, with the maximum time point being t5 = 3.336 × 10. -7 Let n = 6, and begin the operation. The hardware implementation of the operation is as follows: Figure 2 As shown, 30n and -32n are first calculated simultaneously using eight synchronized multipliers. 2 6n 3 、11n 2 -9n, -6n, n 2 The value of 3n is then used to obtain the five coefficients a, b, c, e, and f from the six fixed coefficients through five adders, where coefficient d is a constant and does not need to be calculated; then, the value of j in the waveform characteristic parameter calculation process is obtained from the counter controlled by the data output clock, and a·j is calculated using four multipliers. 2 b·j, d·j 2 The values ​​of e and j are used to obtain [a·j] through two adders. 2 +b·j+c] and [d·j 2 The result is calculated as [e·j+f], which is then multiplied by the original waveform data by a multiplier. The outputs of the two multipliers are then fed into two accumulators to obtain the numerator. With denominator The final calculation result is obtained after passing through the divider.

[0081] The second iteration selects the data from the first iteration and one adjacent data point on each of its left and right sides, i.e., 3.322 × 10. -7 s~3.338×10 -7 The iteration results are recalculated using the eight data points between s.

[0082] Since the data added in this iteration is not between 2 / 3 of the peak voltage (180mV) and the preset threshold (1.296mV), it directly proceeds to the next iteration.

[0083] In subsequent iterations, each time the data used in the calculation was based on the data from the previous iteration, one adjacent data point was added to each side. After multiple iterations, when the newly added data was between 1.296mV and 180mV, the error between the current iteration result and the previous iteration result was less than 3×10⁻⁶. -12 The distance error s, converted to 1 mm, satisfies the iteration termination condition, and the iteration ends. The final measurement result has an error of 0.044 m compared to the actual distance.

[0084] The waveform characteristic parameter calculation method proposed in this invention greatly reduces computational complexity, lowers chip resource consumption and computation time, and the computational complexity does not change dramatically with the increase of data volume. In this example, the entire calculation process is completed directly on the FPGA without the host computer, with a calculation time of only 3.68ms, and the error of the result is similar to that of the full-wave analysis method running on a high-performance computer.

Claims

1. A method for calculating waveform characteristic parameters of a full-waveform lidar, characterized in that, Includes the following steps: Step 1: Set a preset threshold z = η + λ·σ based on the statistical characteristics of the background noise in the echo, where η is the mean of the background noise, σ is the variance of the background noise, and λ is the variance coefficient. Step 2: The lidar emits a pulse signal, and the pulse emission time is recorded as time 0. The receiver digitizes the received echo signal to obtain the original waveform data containing the pulse echo, and determines the maximum value z in the waveform data. max , will z max Extract all data points between the maximum and minimum values ​​at the corresponding time points, and denote them as z. j j = 1, 2, 3, ..., let n be the number of data points extracted, and let t0 be the minimum value at the time corresponding to all the maximum values; Step 3: Calculate the following 6 coefficients based on the number of data points n: Substitute the coefficients and data values ​​into the following calculation formula: The pulse arrival time μ calculated in this operation is obtained; Step 4: Add an original waveform data point, denoted as z, to each side of the data used in the previous iteration. j j = 1, 2, 3, ..., let n be the number of data obtained, let t0 be the minimum value of the data at the corresponding time, repeat step 3 to get the new calculation result μ′ and calculate the difference |μ′-μ| between the result and the previous iteration result; Step 5, if both new data added in step 4 are less than 2*z max If the value of the two new data added in step 4 is less than the preset threshold z, the iteration ends; otherwise, step 4 is repeated. Step 6: Obtain the final feature parameter μ.

2. The method for calculating waveform characteristic parameters of a full-waveform lidar according to claim 1, characterized in that, In step 1, before the lidar emits a pulse signal, the statistical distribution characteristics are obtained from the noise floor data obtained through the analog-to-digital converter. And set a preset threshold based on statistical distribution characteristics.

3. The method for calculating waveform characteristic parameters of a full-waveform lidar according to claim 1, characterized in that, In step 2, the calculation starts from the maximum value of the echo data, and all data between the minimum and maximum values ​​within the time when the maximum value occurs are used as the data for the first iteration.

4. The method for calculating waveform characteristic parameters of a full-waveform lidar according to claim 1, characterized in that, The formula for calculating the coefficient in step 3 is as follows: Based on the characteristics of pulsed lidar echoes, the minimum value in the data involved in the calculation does not affect the relative time difference between the final result and the value. Therefore, setting t0 = 0, the coefficient calculation formula simplifies to: Finally, adding t0 to the calculation result yields the correct feature parameters.

5. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method as described in any one of claims 1-4.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the method as described in any one of claims 1-4.

7. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the steps of the method described in any one of claims 1-4.

Images