KR20230105834A - Lidar apparatus and operating method for the same - Google Patents

Abstract

The present invention relates to a lidar device and an operating method thereof, and by using a lidar sensor to which a SPAD with high light sensitivity is applied, noise caused by external light is corrected and distortion of data due to noise is reduced to increase the detection distance and increase the object detection distance. It improves the sensing performance for , reduces distortion due to noise, can be used in an external environment, and has an effect of improving the accuracy of the measurement distance.

Description

LIDAR device and its operating method {LIDAR APPARATUS AND OPERATING METHOD FOR THE SAME}

The present invention relates to a lidar device that improves sensing performance while reducing noise by using a lidar sensor having high light sensitivity and an operating method thereof.

A lidar sensor detects an object by irradiating light and receiving a signal of the irradiated light reflected from an object, and calculates the distance to the object by measuring the time between a transmission signal and a reception signal.

These lidar sensors have recently been applied to various devices such as automobiles, drones, and robot vacuum cleaners.

In the lidar sensor, a transmitter for irradiating laser light and a receiver for receiving reflected light form a two-dimensional array. Since the lidar sensor continuously operates using the entire array of the transmitter and the receiver that transmit and receive light, power is consumed unnecessarily regardless of the object to be sensed.

The lidar sensor has a problem in that heat is generated as the entire array constituting the transceiver is continuously used.

When the lidar sensor continues to continuously operate the entire array, problems such as acceleration of deterioration of the lidar sensor itself and generation of noise may occur.

In addition, a 2D lidar sensor (LIDAR) constituting a two-dimensional array is effective in object detection and distance calculation using SPAD, but has a problem in that noise interference with internal noise and outdoor light is severe due to high light sensitivity.

In order to solve this problem, a method of measuring received data dozens or hundreds of times to generate a histogram and then detecting a peak value to determine an effective value has been proposed.

However, the lidar sensor composed of SPAD has a problem in that it cannot distinguish whether it is external light or emitted light as external light is incident before the light emitted from the transmitter reaches the receiver, so even if the proposed method is used, an effective value is calculated. There is a problem you can't do.

In addition, the calculation load increases due to the increase in the amount of data to be processed in selecting valid values using the histogram, and there is a problem that it is difficult to apply to small-scale devices.

Republic of Korea Patent Registration No. 10-1877388 describes a technique related to the operation of the LIDAR device.

Republic of Korea Patent Publication No. 10-1877388

The present invention is a lidar device that improves object detection performance by easily processing the amount of data, correcting noise caused by external light, and reducing distortion of data due to noise by using a lidar sensor to which a SPAD with high light sensitivity is applied, and Its purpose is to provide the operation method.

LiDAR device according to the present invention in order to achieve the above object, a lidar sensor (LIDAR) for detecting an object to be sensed by irradiating light and measuring a distance to the object; a learning unit that analyzes the received signal of the lidar sensor to detect a valid value and attenuates noise caused by incident external light; a signal conversion unit measuring a difference between a transmission time and a reception time of a signal in response to the valid value detected by the learning unit; and a control unit calculating a distance to the object based on the difference measured by the signal conversion unit. The learning unit linearizes the received signal of the lidar sensor to reduce the amount of data operation, and calculates the effective value by optimizing a weight based on a difference between a predicted value and an actual value of the linearized data. do.

The learning unit may include: a linearization unit converting index data generated based on the received signal of the lidar sensor into linear data; a difference calculation unit generating a function by calculating a difference between a predicted value and an actual value of the linear data of the linearization unit; and an optimization unit that learns the weights and biases in response to the gradient change rate of the function and optimizes the weights. includes

The learning unit data processing unit for generating a histogram of the received signal of the lidar sensor; a decoding unit that converts the linear data into exponential data based on the weight; and a calculation unit configured to calculate the valid value by detecting a peak value of the histogram from which noise is removed based on the index data. more includes

The linearizer may log the exponential data of the histogram and convert it into linear data.

The linearizer is characterized in that the linearization is performed by applying a logarithmic function to the count values of the histogram using a prediction function.

The difference calculation unit is characterized in that the difference between the predicted value and the actual value is calculated as a least mean square (LMS) and functionalized through a mean square error (MSE).

The optimization unit calculates a gradient change rate of the function to learn weights and biases, and optimizes the gradient by repeating learning until the difference becomes less than a set value.

The decoding unit calculates data of the function using the weights and biases learned in the optimization unit, and refunctionalizes linear data into exponential data based on the calculated data.

The operation unit detects a peak value of the refunctionalized data, and sets data of the signal conversion unit corresponding to the peak value as the effective value.

A method of operating a lidar device according to the present invention includes irradiating light by operating a lidar sensor (LIDAR); Linearizing the received signal of the lidar sensor to reduce the amount of data calculation; Detecting a valid value among the received signals of the lidar sensor by optimizing a weight based on a difference between a predicted value and an actual value of the linearized data; Attenuating noise caused by external light among the received signals of the lidar sensor based on the effective value; measuring a difference between a transmission time and a reception time of a signal corresponding to the effective value; and calculating a distance to the object based on the difference between the transmission time and the reception time. includes

In the case of linearizing the received signal of the lidar sensor, it is characterized in that the index data of the histogram generated based on the received signal of the lidar sensor is logarithmic and converted into linear data.

In the case of linearizing the received signal of the lidar sensor, it is characterized in that the linearization is performed by applying a logarithmic function to the count value of the histogram using a prediction function.

The detecting of the effective value may include calculating a difference between a predicted value and an actual value of the linearized data; generating a function corresponding to the difference; and learning weights and biases in response to the gradient change rate of the function. includes

The step of detecting the valid value may include: optimizing the weight by repeating the step of calculating the difference through machine learning to the step of learning until the difference becomes less than a set value; more includes

The detecting of the effective value may include re-functionalizing linear data by converting the linear data into exponential data based on the weight; and calculating the effective value by detecting a peak value of the histogram from which noise is removed based on the index data.

According to one aspect, the lidar device and its operating method of the present invention have the effect of increasing the detection distance and improving the sensing performance for an object by using a lidar sensor to which a SPAD with high light sensitivity is applied.

According to one aspect of the present invention, it can be used in an external environment by attenuating noise generated due to high light sensitivity and reducing distortion due to noise, and has an effect of improving accuracy of measurement distance.

The present invention has an effect of reducing a large amount of data due to histogram processing through learning to reduce computational load and improving efficiency through fast processing.

1 is a diagram showing the configuration of a lidar device according to an embodiment of the present invention.
2 is a block diagram showing the configuration of the learning unit of FIG. 1;
3 is a flowchart illustrating a flow according to data processing of a lidar device according to an embodiment of the present invention.
4 is a diagram referenced to explain data simulation of a lidar device according to an embodiment of the present invention.
5 is a diagram referenced to explain a method for calculating a difference value using a predictive function of a lidar device according to an embodiment of the present invention.
6 is a diagram referenced to explain a weight optimization method of a lidar device according to an embodiment of the present invention.
7 is a reference diagram for explaining a decoding method for a predictive function of a LIDAR device according to an embodiment of the present invention.
8 is a reference diagram for explaining noise data simulation of a lidar device according to an embodiment of the present invention.
9 is a diagram illustrating simulation results of noise data of a lidar device according to an embodiment of the present invention.
10 is a reference diagram for explaining a method of operating a lidar device according to an embodiment of the present invention.

Hereinafter, the present invention will be described with reference to the accompanying drawings.

In this process, the thickness of lines or the size of components shown in the drawings may be exaggerated for clarity and convenience of explanation. In addition, terms to be described later are terms defined in consideration of functions in the present invention, which may vary according to the intention or custom of a user or operator. Therefore, definitions of these terms will have to be made based on the content throughout this specification.

1 is a diagram showing the configuration of a lidar device according to an embodiment of the present invention.

As shown in FIG. 1, the lidar device includes a lidar sensor 140, a driving unit 120, a signal conversion unit 130, a learning unit 150, a sensor unit 160, a data unit 180, and a control unit 110.

The LIDAR sensor (Light Detection And Ranging) 140 irradiates light to the object 50 to be sensed, detects the presence or absence of the object 50, and measures a distance to the object.

The lidar sensor (LIDAR) 140 is composed of a transmitter 141 and a receiver 142, transmits light through the transmitter 141, and when the light reflected by the object 50 is incident to the receiver 142 , It is configured to detect the presence or absence of the object 50 using this, and to measure the distance using the time when light is reflected and returned.

The lidar sensor (LIDAR) 140 is a 2D lidar (LIDAR) composed of SPAD (Single Photon Avalanche Diode).

The transmitting unit 141 and the receiving unit 142 are each formed in an NxM array structure.

The driving unit 120 controls driving of the transmitting unit 141 and the receiving unit 142 of the lidar sensor 140 in response to a control command of the control unit 110 . The driving unit 120 causes some or all of the arrays of the transmitting unit 141 and the receiving unit 142 of the lidar sensor 140 to operate.

When the operation mode of the lidar sensor 140 is set by the control unit 110, the driving unit 120 controls the array of the transmitting unit 141 and the receiving unit 142 of the lidar sensor 140 to operate in response thereto. .

The driver 120 controls operations of a plurality of light sources of the transmitter 141 and a plurality of photodiodes of the receiver 142 .

In some cases, the driver 120 may control at least some of the plurality of light sources of the transmitter 141 and the plurality of photodiodes of the receiver 142 to operate.

The signal conversion unit 130 may convert a signal in a Time to Digital Converter (TDC) method and also convert a signal in an Analog to Digital Converter (ADC) method. TDC measures the difference between transmission time and reception time, and ADC measures the signal strength.

The signal conversion unit 130 converts the signal by applying any one of TDC and ADC.

The learning unit 150 performs learning by processing data based on machine learning.

The learning unit 150 processes the signal output from the lidar sensor 140 and applies it to the signal conversion unit 130 .

The learning unit 150 linearizes the signal of the lidar sensor, calculates the difference from the actual value, and repeats optimization to re-function (decode).

The learning unit 150 processes the data of the lidar sensor using the SPAD based on machine learning to attenuate noise caused by external light and reduces the amount of data, and applies the data to the signal conversion unit 130.

The data unit 180 stores data of the lidar sensor (LIDAR) 140, learning data according to machine learning-based data processing of the learning unit 150, and data for motion control.

The data unit 180 includes at least one storage unit including a flash memory, and in some cases, a main memory (not shown) for storing the data of the lidar device and a sub-store for storing the data of the learning unit 150. A memory (not shown) may be included.

The sensor unit 160 includes an illuminance sensor. The sensor unit 160 may include a plurality of sensors to sense temperature, humidity, voltage, and current.

The illuminance sensor detects ambient brightness and inputs a signal. The illuminance sensor detects external light and inputs a signal.

The control unit 110 sets an operation mode for the lidar sensor 140, applies a control command to the driving unit 120, and based on the data input from the signal conversion unit 130, the presence or absence of an object and the distance to the object. yields

The control unit 110 stores the data of the LIDAR sensor 140 to which the SPAD is applied and the data of the learning unit 150 learned in the process of attenuating the noise of the data in the data unit 180.

The control unit 110 applies data input by the sensor unit 160 to the learning unit 150 to provide information about the surrounding environment.

When the difference (Δt) between the transmission time and the reception time, that is, the required time is input from the signal conversion unit 130, the control unit 110 calculates the distance to the object based on this.

The control unit 110 may repeatedly calculate the distance at regular time intervals corresponding to the movement of the lidar device or object, and output data therefor.

For example, when a lidar device is installed in a vehicle, the controller 110 may transmit information about the distance to the object to the main controller of the vehicle.

2 is a block diagram showing the configuration of the learning unit of FIG. 1;

As shown in FIG. 2, the learning unit 150 includes a data processing unit 151, a linearization unit 152, a difference calculation unit 153, an optimization unit 154, a calculation unit 155, and a decoding unit 156. include

The data processing unit 151 processes data of the lidar sensor (LIDAR) 140 according to the control command of the control unit 110.

The data processing unit 151 generates a histogram based on the received signal of the lidar sensor (LIDAR) 140 and post-processes the data.

When an external light detection signal is input from the illuminance sensor of the sensor unit 160 in addition to the data of the LIDAR sensor 140, the data processing unit 151 receives and processes it.

The data processing unit 151 converts the data of the lidar sensor (LIDAR) 140 into exponential data to generate a histogram.

The linearizer 152 converts exponential data into linear data for machine learning.

The linearization unit 152 transforms the exponential data into logarithmic data. The linearization unit 152 converts ln into linear data for the y value that is the histogram count value (count) for fitting with the predictive function.

The difference calculation unit 153 calculates the difference between the predicted value and the actual value. The difference calculation unit 153 generates a cost function for the difference between the predicted value and the actual value.

The difference calculation unit 153 calculates the difference from the actual data based on the prediction function and performs functionalization through mean square error (MSE).

The optimizer 154 calculates a gradient change rate of the function with respect to the cost function calculated by the difference value calculator 153 and learns weights and biases.

The optimization unit 154 feeds back the calculated weight and bias to the difference value calculation unit 153, and calculates the difference value until the difference between the actual value and the predicted value calculated by the difference value calculation unit 153 is reduced to less than a certain value. Repeat learning about weights and biases.

The optimizer 154 optimizes weights by iteratively learning weights and biases.

The decoding unit 156 derives a value of an existing prediction function through the weight (W) and the bias (b) learned in the optimization unit 154.

The decoding unit 156 derives j and k values from the predictive function of Equation 1 described later.

The calculation unit 155 distinguishes between noise caused by external light and a sensor signal among signals of the LIDAR sensor 140 .

The calculator 155 calculates the difference between the histogram data and the predicted value, and detects a peak value for the difference. The calculation unit 155 determines the data value of the signal conversion unit 130 corresponding to the detected peak value as a valid value.

The operation unit 155 determines the transmission time and reception time corresponding to the peak value and the difference between the transmission time and the reception time of the signal conversion unit 130 as effective values, thereby distinguishing external light from the sensor signal. can

The calculation unit 155 removes noise from histogram data by filtering a signal caused by external light.

The data processing unit 151 applies the noise-removed data to the signal conversion unit 130 .

3 is a flowchart illustrating a flow according to data processing of a lidar device according to an embodiment of the present invention.

As shown in FIG. 3, when a control command is applied from the control unit 110 to operate the lidar sensor (LIDAR) 140 (S110), the driving unit 120 drives the lidar sensor (LIDAR) 140. It does (S120).

The lidar sensor (LIDAR) 140 irradiates light through a light source provided in the transmitter 141, and the irradiated light is reflected on an object and incident to the receiver 142.

The learning unit 150 receives the signal of the lidar sensor (LIDAR) 140, generates a histogram, and removes noise by performing linearization and optimization based on machine learning.

Since external light in addition to the signal (light) transmitted from the transmitter 141 is incident on the receiver 142 of the LIDAR sensor 140, the learning unit 150 removes noise caused by the external light.

The data processing unit 151 of the learning unit 150 generates a histogram based on the data of the LIDAR sensor 140 (S130) and post-processes the data (S150).

When external light is detected by the illuminance sensor of the sensor unit 160, the data processing unit 151 reflects the external light and performs data post-processing.

The data processing unit 151 applies data to the linearization unit 152 when an interrupt occurs (S160).

The linearizer 152 linearizes the histogram data (S170). The linearization unit 152 converts the histogram exponential data into logarithmic data and linearizes it. The linearization unit 152 converts the y value, which is the count value of the histogram, into logarithmic data for fitting with the predictive function.

The linearizer 152 filters the data through a filter in the process of linearizing the data (S180).

The difference calculation unit 153 receives the linear data of the linearization unit 152 and performs a least mean square (LMS) operation (S190). The difference value calculation unit 153 reduces the difference by selecting a value that minimizes the sum of squares of the difference between the predicted value and the actual value through least mean square operation.

The difference calculation unit 153 calculates the difference between the predicted value and the actual value, and performs functionalization through mean square error (MSE) to generate a cost function.

The optimization unit 154 optimizes the gradient by learning weights and biases by calculating a gradient change rate of the function with respect to the cost function calculated by the difference calculation unit 153 (S200).

The optimization unit 154 feeds back the calculated weight and bias to the difference value calculation unit 153, and calculates the difference value until the difference between the actual value and the predicted value calculated by the difference value calculation unit 153 decreases to less than the set value. Repeat learning about weights and biases.

The decoding unit 156 derives a value of an existing prediction function through the weight (W) and the bias (b) learned in the optimization unit 154 and re-functions it (S210).

The calculation unit 155 performs data filtering (S220) and removes noise (S230).

The operation unit 155 calculates the mode value from the noise-removed histogram (S240), detects the transmission time, reception time, and difference of the LIDAR sensor corresponding to the mode and applies them to the signal conversion unit 130 (S250). . The arithmetic unit 155 distinguishes between noise caused by external light and a sensor signal among signals of the LIDAR sensor 140 to remove noise caused by external light.

The signal conversion unit 130 converts the data corresponding to the most frequent value detected by the operation unit 155 into a signal, and the control unit 110 calculates the distance to the object based on the data of the signal conversion unit 130 .

4 is a diagram referenced to explain data simulation of a lidar device according to an embodiment of the present invention.

The linearization unit 152 converts the exponential data into linear data before the optimization unit 154 performs machine learning.

As shown in (a) of FIG. 4, the linearization unit 152 converts the exponential data of the histogram data into log data and converts it into linear data, and for fitting with the prediction function, the histogram count value (count) Take ln to the y value and convert it to linear data.

As shown in (b) of FIG. 4, the linearization unit 152 converts exponential data into linear data.

The prediction function is as shown in Equation 1 below.

5 is a diagram referenced to explain a method for calculating a difference value using a predictive function of a lidar device according to an embodiment of the present invention.

As shown in FIG. 5 , the difference calculation unit 153 calculates the difference from actual data based on the prediction function described above and performs functionalization through mean square error (MSE).

The difference calculation unit 153 calculates a cost function by functionalizing the difference from actual data as shown in Equation 2 below.

W is the weight and b is the bias.

The cost function uses w1 as the starting point (P1), literates each section (L3, L4), and has a convergence value of the final value (wn).

6 is a diagram referenced to explain a weight optimization method of a lidar device according to an embodiment of the present invention.

The optimizer 154 calculates a gradient change rate of the function with respect to the cost function calculated by the difference value calculator 153 and learns weights and biases.

The optimizer 154 learns weights and biases as shown in Equation 3.

W is the weight and b is the bias.

As shown in FIG. 6 , the optimizer 154 optimizes weights by iteratively learning weights and biases.

The initially distributed weight values are optimized to a single value through machine learning.

The optimizer 154 calculates the gradient change rate of the cost function, and learns the weight and bias towards the inflection point in response to the gradient change rate. In the case of a quadratic function having a positive quadratic coefficient, the optimizer 154 sets the minimum value of the slope change rate as the inflection point C1 to learn weight and bias.

7 is a reference diagram for explaining a decoding method for a predictive function of a LIDAR device according to an embodiment of the present invention.

The decoding unit 156 derives a value of an existing prediction function through the weight (W) and the bias (b) learned in the optimization unit 154.

As shown in (a) of FIG. 7, the decoding unit 156 applies the weight and bias optimized by the optimization unit 154 to the function of x and ln(y) to obtain j in the prediction function of Equation 1, Derive k value.

As shown in (b) of FIG. 7, the decoding unit 156 converts the linearized data into exponential data and re-functions it.

8 is a reference diagram for explaining noise data simulation of a lidar device according to an embodiment of the present invention.

As shown in (a) of FIG. 8 , data L11 distorted due to noise caused by external light or the like is included in the histogram.

The linearization unit 152 converts the histogram exponential data into log data and converts it into linear data as shown in FIG. 8(b).

By converting into linear data in the linearization unit 152, the amount of data calculation is reduced.

The difference calculation unit 153 calculates the difference between the actual value and the predicted value, and the optimization unit 154 optimizes the weights as shown in FIG. 6 by learning weights and biases based on the difference values.

9 is a diagram illustrating simulation results of noise data of a lidar device according to an embodiment of the present invention.

As shown in (a) of FIG. 9 , when the weights are optimized by the optimization unit 154, the decoding unit 156 applies them to the prediction function to calculate data of the prediction function.

The decoding unit 156 converts the linearized data back into exponential data based on the data of the prediction function.

The decoding unit 156 outputs a histogram from which noise is removed through weight optimization.

As shown in (b) of FIG. 9 , the calculation unit 155 detects a peak value from the histogram data and determines a TCD value corresponding to the peak value as a valid value. Accordingly, the arithmetic unit 155 corrects the sensor data by separating the noise caused by external light and the signal of the sensor among the signals of the LIDAR sensor and removing the noise (L12, L13).

The calculation unit 155 removes noise from histogram data by filtering a signal caused by external light.

10 is a reference diagram for explaining a method of operating a lidar device according to an embodiment of the present invention.

As shown in FIG. 10, the lidar sensor (LIDAR) 140 operates, and a plurality of light sources of the transmitter 141 emit light (S310). The irradiated light is reflected on the object, and the reflected light is incident to the receiver 142 of the LIDAR sensor 140. At this time, external light as well as the transmitted light may be incident to the receiving unit 142 of the LIDAR sensor 140 and act as noise.

The learning unit 150 processes the received data of the lidar sensor (LIDAR) 140 based on machine learning to distinguish external light from a signal and remove noise.

The data processing unit 151 of the learning unit 150 generates the data of the lidar sensor (LIDAR) 140 as a histogram (S320) and post-processes the data (S330).

The linearization unit 152 converts the histogram index data into linear data by logarithmic data (S340). The linearization unit 152 linearizes the count values of the histogram by applying a logarithmic function using the prediction function.

The difference calculation unit 153 calculates the difference between the predicted value and the actual value in the predictive function with respect to the linear data. The difference value calculation unit 153 calculates the difference value through least mean square (LMS) operation (S350).

The optimization unit 154 repeatedly learns and optimizes weights and biases using the cost function calculated by the difference calculation unit 153 (S360).

The optimization unit 154 feeds back the calculated weight and bias, and the difference calculation unit 153 repeats the calculation until a difference value smaller than the set value is calculated (S370).

The decoding unit 156 derives a value of an existing prediction function through the weight (W) and the bias (b) learned in the optimization unit 154, and converts the linear data into log data (S380).

The operation unit 155 removes the noise converted through filtering and analyzes the histogram data from which the noise has been removed to detect a peak value. The calculation unit 155 determines the data of the lidar sensor corresponding to the peak value as a valid value (S400).

The signal conversion unit 130 converts the data detected by the operation unit 1550 and applies it to the control unit 110, and the control unit 110 determines the distance to the object in response to the difference between the transmission time and the reception time of the applied data. It is calculated (S420).

Accordingly, the present invention variably controls the resolution of the lidar sensor (LIDAR) 140 according to the distance to improve object recognition performance while optimizing the operation of the lidar sensor (LIDAR) 140 and reducing heat and noise. can solve the problem.

The present invention detects an object by applying a SPAD with high light sensitivity, reduces the amount of data calculation through linearization, performs iterative calculation on the difference between the predicted value and the actual value, and optimizes the weight to distinguish external light from the signal of the sensor. Thus, noise caused by external light can be attenuated, and the object can be accurately detected and the distance can be calculated.

Although described with reference to the embodiment shown in the drawings, this is only exemplary, and those skilled in the art will understand that various modifications and equivalent other embodiments are possible therefrom. Therefore, the true technical protection scope of the present invention should be defined by the claims below.

110: control unit 120: driving unit
130: signal conversion unit 140: lidar sensor
141: transmitter 142: receiver
150: learning unit 152: linearization unit
153: difference calculation unit 154: optimization unit
155: calculation unit 156: decoding unit
160: sensor unit 180: data unit

Claims (15)

A lidar sensor (LIDAR) for detecting an object to be sensed by irradiating light and measuring a distance to the object;
a learning unit that analyzes the received signal of the lidar sensor to detect a valid value and attenuates noise caused by incident external light;
a signal conversion unit measuring a difference between a transmission time and a reception time of a signal in response to the valid value detected by the learning unit; and
a control unit for calculating a distance to the object based on the difference measured by the signal conversion unit; including,
The learning unit linearizes the received signal of the lidar sensor to reduce the amount of data operation, and calculates the effective value by optimizing a weight based on a difference between a predicted value and an actual value of the linearized data. . According to claim 1,
The learning unit,
A linearization unit for converting index data generated based on the received signal of the lidar sensor into linear data;
a difference calculation unit generating a function by calculating a difference between a predicted value and an actual value of the linear data of the linearization unit; and
an optimization unit that learns the weights and biases in response to the gradient change rate of the function and optimizes the weights; Lidar device comprising a. According to claim 2,
The learning department
A data processing unit for generating a histogram of the received signal of the lidar sensor;
a decoding unit that converts the linear data into exponential data based on the weight; and
a calculator configured to calculate the effective value by detecting a peak value of a histogram from which noise is removed based on the exponential data; A lidar device further comprising a. According to claim 2,
LiDAR device, characterized in that the linearization unit converts the logarithmic data of the histogram into linear data. According to claim 4,
LiDAR device, characterized in that the linearization unit linearizes by applying a logarithmic function to the count value of the histogram using a prediction function. According to claim 2,
LiDAR device, characterized in that the difference value calculation unit calculates the difference between the predicted value and the actual value as a least mean square (LMS), and functions through a mean square error (MSE). According to claim 2,
The optimization unit calculates the gradient change rate of the function to learn weights and biases, and optimizes the gradient by repeating the learning until the difference is less than the set value. According to claim 3,
LiDAR device, characterized in that the decoding unit calculates the data of the function using the weights and biases learned in the optimization unit, and re-functions linear data into exponential data based on the calculated data. According to claim 3,
LiDAR device, characterized in that the calculation unit detects the peak value of the refunctionalized data, and the data of the signal conversion unit corresponding to the peak value as the effective value. A lidar sensor (LIDAR) operating to irradiate light;
Linearizing the received signal of the lidar sensor to reduce the amount of data calculation;
Detecting a valid value among the received signals of the lidar sensor by optimizing a weight based on a difference between a predicted value and an actual value of the linearized data;
Attenuating noise caused by external light among the received signals of the lidar sensor based on the effective value;
measuring a difference between a transmission time and a reception time of a signal corresponding to the effective value; and
calculating a distance to an object based on a difference between the transmission time and the reception time; Method of operating a lidar device comprising a. According to claim 10,
When linearizing the received signal of the lidar sensor,
Method of operating a lidar device, characterized in that for converting the index data of the histogram generated on the basis of the received signal of the lidar sensor into linear data. According to claim 11,
When linearizing the received signal of the lidar sensor,
Method of operating a lidar device, characterized in that for linearization by applying a log function to the count value of the histogram using a prediction function. According to claim 10,
In the step of detecting the valid value,
calculating a difference between a predicted value and an actual value of the linearized data;
generating a function corresponding to the difference; and
learning weights and biases in response to the gradient change rate of the function; Method of operating a lidar device comprising a. According to claim 13,
In the step of detecting the valid value,
optimizing the weight by repeating the step of calculating the difference through machine learning until the step of learning is less than the set value; Method of operating a lidar device further comprising a. According to claim 13,
In the step of detecting the valid value,
Based on the weight, converting linear data into exponential data and re-functionalizing it; and
Based on the index data, a method of operating a lidar device comprising the step of calculating the effective value by detecting the peak value of the histogram from which noise is removed.

Images