Research On Key Technology Of High Performance LiDAR Detection System

At present,autonomous vehicles based on cameras or millimeter waves can recognize and process conventional scenes.However,autonomous vehicles based on cameras or millimeter waves will fail to handle emergency.To deal with unexpected application scenarios,it is very necessary to introduce the light detection and ranging(Li DAR)system in autonomous vehicles.The traditional Li DAR system has three insufficient.Firstly,the system uses single point avalanche photodiode as photoelectric detection that makes the system large in size and power consumption.Secondly,the system needs to use digital signal processor(DSP)to achieve the post-processing algorithm to reduce the imaging efficiency of the system.Finally,the system uses fixed encoding method which limits the widespread popularity of Li DAR systems.In order to solve the existing problems of the Li DAR system,this thesis is used to investigate a prototype of Li DAR system with low power consumption,high accuracy,and high imaging frame rate.The thesis proposes a high-power,phase-modulated multi-pulse laser drive circuit.The time interval between multiple pulsed laser signals obeys pseudo-random distribution.The pseudo-random digital code is generated based on FPGA,which is used to convert into the corresponding time interval using the digital to time converter(DTC).The pulse generation module driven by DTC generates two pulse signals that is used to drive the laser driver.The time interval between the pulse signals is proportion to the output of DTC.The coding method of the emitted laser signal between different Li DAR system is unique that can eliminate the interference of inter-Li DAR system and improve ability of suppression of background light interference.The probability of Li DAR system being falsely triggered is less than one percent When the background light intensity is less than 70 klux.The thesis investigates a high-performance Li DAR receiving system.The sixteen channel avalanche photo diode(APD)is used as the photoelectric detection.The trans-impedance amplifier(TIA)with eight channel and sixteen channel amplifier circuit are used to convert the pulse current of the APD output into pulse voltage and amplify it.The time discrimination circuit is used to convert the analog pulse voltage signal into digital signal.The temperature sensor and the analog to digital converter(ADC)are used to sample the temperature information on the receiving board in real time which is used to control system to adjust the driving voltage of the laser driver and APD to ensure that the gain of APD remains the same,output power and wavelength of the laser diode(LD)are basically unchanged.The system can work well when the temperature ranging is in-10 degrees Celsius to 55 degrees Celsius.The thesis investigates the multi-channel time to digital converter(TDC)based on field programmable gate array(FPGA),which can be used to measure multi-time interval at the same time.TDC combines of coarse-counter and fine-counter.To reduce the consumption of the logic resources,one TDC shares a coarse counter and a delay unit.The proposed TDC can measure multi-time intervals and improve the measurement efficiency without increasing the resources and reducing the measurement accuracy.The code density measurement is implemented based on the FPGA that is used to adjust the delay time of each delay unit in real time to reduce the influence of temperature and power jitter on the TDC measurement accuracy.The weight filter is realized in FPGA to improve the measurement precision of the TDC.The measurement results show that the measurement accuracy and precision of the proposed TDC are less than 300 ps and 100 ps.The thesis proposes an equivalent sampling circuit based on time discrimination circuit,digital to analog converter(DAC)and TDC.Based on the sampling data,the system uses hybrid amplitude fitting algorithm to calculate the amplitude of echo signal.The hybrid amplitude fitting algorithm consists of Gaussian fitting algorithm,exponential fitting algorithm,and linear fitting algorithm.The system can dynamically select the appropriate fitting algorithm according to the actual measurement results to improve the measurement precision.When the analog circuit is not saturated,the system obtains the target reflectivity using the Gaussian fitting algorithm.When the analog circuit is Undersaturated,the system obtains the target reflectivity using exponential fitting algorithm.When the analog circuit is in the supersaturated,the system uses the linear fitting algorithm to obtain the target reflectivity.The thesis proposes a background light suppression algorithm to resist stray light interference and cross relation fitting algorithm(CSFA)to reduce the walk error to improve measurement accuracy.The CSFA is an improved algorithm for solving nonlinear solutions,which divides the effective fitting interval into multiple equally time intervals.Within the same interval,the linear fitting algorithm is used to calculate the nonlinear equations.And the slope between adjacent intervals satisfies the distribution of the Gaussian function.Compared with the traditional walk error calibration algorithm,CSFA does not need to obtain the compensation coefficient based on experiments in advance.A large number of experiments indoor and outdoor were conducted using the developed Li DAR system to evaluate the measurement accuracy,measurement precision,spatial resolution,and anti-stray light interference ability.The experimental results show that the developed Li DAR system can achieve an imaging frame rate of 10 frames per second with a resolution of 16×128 pixels.The measurement range accuracy of the system is ±1.5cm when measurement ranging is less than 20 m.The relative measurement error is less than0.1 percent when the measurement ranging is greater than 20 m.The measurement ranging is 0.4m to 100 m.