Passenger car point cloud map generation method and system based on simulation engine

Abstract

The invention relates to a passenger vehicle point cloud map generation method and system based on a simulation engine, and the method comprises the steps: U1, carrying out the integration and optimization based on the data information of a high-precision map, a dynamic model of a vehicle, and a simulation engine through employing a multi-mode integrated optimization algorithm based on a refraction reverse learning mechanism and a self-adaptive control factor, constructing a high-precision simulation environment of the vehicle; and U2, based on the high-precision simulation environment of the vehicle, inputting the data information of the parameters of the vehicle controller and the data information of the parameters of the positioning controller of the vehicle to perform analogue simulation of the vehicle, and outputting the data information of the motion state of the vehicle, the data information of the sensor and the data information of positioning. The precision and compatibility of a simulation system are improved, high-frequency positioning signals and Lidar point cloud data can be simulated, the requirement for point cloud map generation is met, a new and efficient method is provided for passenger car point cloud map generation, and the production cost and period are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of point cloud map generation, and in particular to a method and system for generating a point cloud map of a passenger car based on a simulation engine. Background Art

[0002] With the rapid development of autonomous driving technology, point cloud maps play an increasingly important role in the navigation, positioning and environmental perception of autonomous vehicles. Point cloud maps are a type of map composed of a large amount of three-dimensional point cloud data that can accurately represent the shape, position and posture of objects, providing accurate environmental information for autonomous vehicles. However, the production process of point cloud maps is complex and time-consuming, requiring high-precision sensor data and complex processing algorithms. Therefore, how to improve the production efficiency and accuracy of point cloud maps has become one of the current research hotspots.

[0003] In the prior art one, simulation testing of vehicle control systems is usually achieved based on real-time simulation systems such as VT System, combined with modular hardware platforms. These simulation systems can simulate the dynamic characteristics of the vehicle, sensor data, etc., but are often limited by the performance of hardware and software and cannot simulate high-frequency positioning signals and complex road environments. Due to the relatively simple circuit structure, it is impossible to simulate high-frequency positioning signals, resulting in the positioning algorithm being unable to effectively identify the positioning data information sent by the IMU in the simulation environment. In addition, the prior art one lacks dedicated simulation modules and tools for point cloud map production, and cannot generate high-quality point cloud map data.

[0004] MATLAB / Simulink is a software environment developed by MathWorks for algorithm development, simulation, automatic code generation and embedded system design. In the second prior art, MATLAB / Simulink is usually used in combination with a custom hardware platform to realize simulation testing of automobile control systems. These simulation software have rich functions and powerful performance, and can simulate complex control algorithms and sensor data. First, the development cost of custom hardware platforms is high and the cycle is long, which is not conducive to rapid iteration and optimization; secondly, the integration of MATLAB / Simulink and custom hardware platforms is not high, resulting in delays in data transmission and processing, affecting simulation efficiency and accuracy; finally, there is a lack of dedicated simulation modules and tools for point cloud map production, and it is impossible to accurately simulate the point cloud data output of sensors such as Lidar, resulting in deviations between simulation results and actual road environments.

[0005] In recent years, with the rise of open source technology, some open source simulation engines (such as Unity, Unreal Engine, etc.) have gradually been applied to the simulation test of automotive control systems. These open source simulation engines have rich graphics rendering capabilities, physical simulation capabilities and AI algorithm support, and can simulate real-world road and traffic conditions. Although open source simulation engines have excellent performance in terms of functions and performance, there are still some challenges in the simulation test of automotive control systems. First, the integration and optimization of open source simulation engines requires certain technical strength and experience; second, due to the openness and flexibility of open source simulation engines, they may face risks in terms of security and stability; finally, in terms of point cloud map production, the dedicated modules and tools of open source simulation engines are still limited and need further development and optimization. Summary of the invention

[0006] In view of the above deficiencies in the prior art, the present invention provides a method and system for generating point cloud maps for passenger cars based on a simulation engine, which not only improves the accuracy and compatibility of the simulation system, can simulate high-frequency positioning signals and Lidar point cloud data, and meet the needs of point cloud map generation, but also provides a new and efficient method for generating point cloud maps for passenger cars, thereby reducing production costs and cycles.

[0007] In order to achieve the above-mentioned purpose and other related purposes, the technical solution provided by the present invention is as follows:

[0008] A method for generating a point cloud map of a passenger car based on a simulation engine, the method comprising:

[0009] U1. Based on the data information of high-precision maps, the dynamic model of the vehicle and the simulation engine, a multi-modal integrated optimization algorithm based on refractive reverse learning mechanism and adaptive control factor is used for integration and optimization to build a high-precision simulation environment for the vehicle;

[0010] U2. Based on the high-precision simulation environment of the vehicle, the data information of the parameters of the vehicle controller and the data information of the parameters of the vehicle positioning controller are input to simulate the vehicle, and the data information of the vehicle's motion state, the data information of the sensor and the positioning data information are output;

[0011] U3. Based on the data information of the motion state of the vehicle, the data information of the sensor and the positioning data information, a polynomial feature extraction algorithm based on chaotic mapping is used to extract the geometric information of the road, the position of the traffic sign and the information of the obstacle distribution, and obtain the geometric information of the road, the position of the traffic sign and the information of the obstacle distribution;

[0012] U4. Based on the geometric information of the road, the location of traffic signs and the information of obstacle distribution, a point cloud map generation model of the road is constructed, and the point cloud map of the road is constructed to obtain data information of the point cloud map of the road.

[0013] Furthermore, in step U1, the integration and optimization using a multi-modal integrated optimization algorithm based on a refraction reverse learning mechanism and an adaptive control factor includes:

[0014] U1. Based on the data information of the high-precision map and the dynamic model of the vehicle, an interaction function Q based on the refraction reverse learning mechanism is established.

[0015]

[0016] Among them, x is the data information of the high-precision map, y is the dynamic model of the vehicle, and α 1 , α 2 and α 3 is the interactive learning factor, f is the refraction reverse learning mechanism function, which characterizes the interactive information between the vehicle and the high-precision map, and obtains the interactive data information between the vehicle and the high-precision map;

[0017] U12. Based on the data information of the interaction between the vehicle and the high-precision map and the simulation engine, a multimodal integrated function W based on an adaptive control factor is established.

[0018]

[0019] Among them, r is the data information of the interaction between the vehicle and the high-precision map, h is the simulation engine, and β 1 , β 2 and β 3 is the adaptive control factor;

[0020] U13. Based on the multi-modal integrated function W based on the adaptive control factor, integration and optimization are performed to build a high-precision simulation environment for the vehicle.

[0021] Furthermore, the adaptive control factor β 1 , β 2 and β 3 The constraint function g is,

[0022] Among them, the value range of the constraint function g is (0,1).

[0023] Furthermore, the interactive learning factor α 1 , α 2 and α 3 The constraints are:

[0024]

[0025] The refraction reverse learning mechanism function f is,

[0026]

[0027] Among them, y is the dynamic model of the vehicle.

[0028] Furthermore, the simulation engine includes an IMU data simulation model and a laser point cloud data simulation model, wherein the IMU data simulation model establishes an IMU data simulation function R according to the data information of the parameters of the vehicle controller and the data information of the parameters of the vehicle positioning controller.

[0029]

[0030] Among them, z 1 is the data information of the parameters of the vehicle controller, z 2 is the data information of the parameters of the vehicle positioning controller, δ 1 , δ 2 and δ 3 is the error parameter factor of IMU data simulation, and simulates the IMU positioning data of the vehicle; the laser point cloud data simulation model is to establish a laser point cloud data simulation function P according to the data information of the parameters of the vehicle controller and the data information of the parameters of the vehicle positioning controller,

[0031]

[0032] Among them, z 1 is the data information of the parameters of the vehicle controller, z 2 The data information of the parameters of the vehicle positioning controller, γ 1 , γ 2 and γ 3 The error factor of laser point cloud data simulation is used to simulate the laser point cloud data of the vehicle.

[0033] Furthermore, in step U3, the extraction of road geometry information, traffic sign locations, and obstacle distribution information using a polynomial feature extraction algorithm based on chaotic mapping includes:

[0034] U31. Based on the data information of the vehicle's motion state, the sensor data information and the positioning data information, establish the vehicle's chaotic mapping function S,

[0035]

[0036] Among them, a 1 is the data information of the vehicle's motion status, a 2is the data information of the vehicle sensor, a 3 is the vehicle positioning data information, η 1 , η 2 and η 3 is the weight factor of the chaotic mapping, and the vehicle data is fused to obtain the fused vehicle data information;

[0037] U32. Based on the data information of the fused vehicle, a polynomial feature extraction function F of the vehicle is established.

[0038]

[0039] Among them, b is the data information of the vehicle after fusion processing, λ 0 , 1 , 2 and λ 3 Extracting factors for the polynomial features of the vehicle;

[0040] U33. Based on the polynomial feature extraction function F of the vehicle, the geometric information of the road, the position of the traffic sign and the information of the obstacle distribution are extracted to obtain the geometric information of the road, the position of the traffic sign and the information of the obstacle distribution.

[0041] Furthermore, the polynomial feature extraction factor λ of the vehicle 0 , 1 , 2 and λ 3 The constraints are:

[0042]

[0043] Furthermore, in step U4, the step of constructing a point cloud map generation model for a road includes:

[0044] U41. Based on the geometric information of the road, the location of traffic signs and the information of obstacle distribution, data cleaning, denoising and calibration are performed to obtain the processed geometric information of the road, the location of traffic signs and the information of obstacle distribution;

[0045] U42. Based on the processed road geometry information, traffic sign location and obstacle distribution information, establish a road point cloud map generation function G,

[0046]

[0047] Among them, c 1 is the geometric information of the processed road, c 2 is the position of the traffic sign after processing, c 3 is the obstacle distribution information after processing, μ1 , μ 2 and μ 3 Determinants of point cloud map generation;

[0048] U43. Based on the point cloud map generation function G of the road, the point cloud map of the road is constructed to obtain data information of the point cloud map of the road.

[0049] Furthermore, the determining factor μ for generating the point cloud map is 1 , μ 2 and μ 3 The constraint function L is,

[0050]

[0051] Among them, the value range of the constraint function L is (3,5).

[0052] In order to achieve the above-mentioned purpose and other related purposes, the present invention also provides a passenger car point cloud map generation system based on a simulation engine, including a computer device, which is programmed or configured to execute any one of the steps of the passenger car point cloud map generation method based on a simulation engine.

[0053] The present invention has the following positive effects:

[0054] 1. The present invention integrates and optimizes a multimodal integrated optimization algorithm based on a refractive reverse learning mechanism and an adaptive control factor to construct a high-precision simulation environment for vehicles, and extracts road geometry information, traffic sign locations, and obstacle distribution information using a polynomial feature extraction algorithm based on chaotic mapping. This not only improves the accuracy and compatibility of the simulation system, can simulate high-frequency positioning signals and Lidar point cloud data, meet the needs of point cloud map generation, optimizes the simulation data processing and analysis process, improves the efficiency and accuracy of data processing, but also provides a new and efficient method for generating point cloud maps for passenger cars, reducing production costs and cycles.

[0055] 2. The present invention constructs a point cloud map of the road by constructing a point cloud map generation model of the road, which not only improves the safety and reliability of autonomous driving vehicles, provides strong support for the development of autonomous driving technology, but also promotes the application and development of simulation technology in the field of autonomous driving, and promotes the innovation and progress of autonomous driving technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a schematic diagram of the method flow of the present invention;

[0057] Figure 2It is a flow chart of the multi-modal integrated optimization algorithm based on the refractive reverse learning mechanism and the adaptive control factor of the present invention;

[0058] Figure 3 It is a schematic diagram of the flow of the polynomial feature extraction algorithm based on chaotic mapping of the present invention;

[0059] Figure 4 A schematic diagram of a process for constructing a point cloud map generation model for a road according to the present invention;

[0060] Figure 5 It is a schematic diagram of the simulation engine framework of the present invention;

[0061] Figure 6 It is a schematic diagram of the positioning signal and Lidar point cloud data simulation module of the present invention;

[0062] Figure 7 This is a schematic diagram of laser point cloud simulation modeling of the present invention;

[0063] Figure 8 The figure is a flow chart of the simulated radar data matching process of the present invention. DETAILED DESCRIPTION

[0064] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0065] Example 1: Figure 1 As shown, a method for generating a point cloud map of a passenger car based on a simulation engine, the method comprising:

[0066] U1. Based on the data information of high-precision maps, the dynamic model of the vehicle and the simulation engine, a multi-modal integrated optimization algorithm based on refractive reverse learning mechanism and adaptive control factor is used for integration and optimization to build a high-precision simulation environment for the vehicle;

[0067] U2. Based on the high-precision simulation environment of the vehicle, the data information of the parameters of the vehicle controller and the data information of the parameters of the vehicle positioning controller are input to simulate the vehicle, and the data information of the vehicle's motion state, the data information of the sensor and the positioning data information are output;

[0068] U3. Based on the data information of the motion state of the vehicle, the data information of the sensor and the positioning data information, a polynomial feature extraction algorithm based on chaotic mapping is used to extract the geometric information of the road, the position of the traffic sign and the information of the obstacle distribution, and obtain the geometric information of the road, the position of the traffic sign and the information of the obstacle distribution;

[0069] U4. Based on the geometric information of the road, the location of traffic signs and the information of obstacle distribution, a point cloud map generation model of the road is constructed, and the point cloud map of the road is constructed to obtain data information of the point cloud map of the road.

[0070] In this embodiment, if Figure 2 As shown, in step U1, the integration and optimization using a multi-modal integrated optimization algorithm based on a refraction reverse learning mechanism and an adaptive control factor includes:

[0071] U1. Based on the data information of the high-precision map and the dynamic model of the vehicle, an interaction function Q based on the refraction reverse learning mechanism is established.

[0072]

[0073] Among them, x is the data information of the high-precision map, y is the dynamic model of the vehicle, and α 1 , α 2 and α 3 is the interactive learning factor, f is the refraction reverse learning mechanism function, which characterizes the interactive information between the vehicle and the high-precision map, and obtains the interactive data information between the vehicle and the high-precision map;

[0074] U12. Based on the data information of the interaction between the vehicle and the high-precision map and the simulation engine, a multimodal integrated function W based on an adaptive control factor is established.

[0075]

[0076] Among them, r is the data information of the interaction between the vehicle and the high-precision map, h is the simulation engine, and β 1 , β 2 and β 3 is the adaptive control factor;

[0077] U13. Based on the multi-modal integrated function W based on the adaptive control factor, integration and optimization are performed to build a high-precision simulation environment for the vehicle.

[0078] In this embodiment, the adaptive control factor β 1 , β 2 and β 3 The constraint function g is,

[0079]

[0080] Among them, the value range of the constraint function g is (0,1).

[0081] In this embodiment, the interactive learning factor α 1 , α 2 and α 3 The constraints are:

[0082] The refraction reverse learning mechanism function f is,

[0083]

[0084] Among them, y is the dynamic model of the vehicle.

[0085] In this embodiment, the simulation engine includes an IMU data simulation model and a laser point cloud data simulation model. The IMU data simulation model establishes an IMU data simulation function R according to the data information of the parameters of the vehicle controller and the data information of the parameters of the vehicle positioning controller.

[0086]

[0087] Among them, z 1 is the data information of the parameters of the vehicle controller, z 2 is the data information of the parameters of the vehicle positioning controller, δ 1 ,δ 2 and δ 3 is the error parameter factor of IMU data simulation, and simulates the IMU positioning data of the vehicle; the laser point cloud data simulation model is to establish a laser point cloud data simulation function P according to the data information of the parameters of the vehicle controller and the data information of the parameters of the vehicle positioning controller,

[0088]

[0089] Among them, z 1 is the data information of the parameters of the vehicle controller, z 2 The data information of the parameters of the vehicle positioning controller, γ 1 , γ 2 and γ 3 The error factor of laser point cloud data simulation is used to simulate the laser point cloud data of the vehicle.

[0090] In this embodiment, if Figure 3 As shown, in step U3, the use of a polynomial feature extraction algorithm based on chaotic mapping to extract road geometry information, traffic sign locations, and obstacle distribution information includes:

[0091] U31. Based on the data information of the vehicle's motion state, the sensor data information and the positioning data information, establish the vehicle's chaotic mapping function S,

[0092]

[0093] Among them, a 1 is the data information of the vehicle's motion status, a 2 is the data information of the vehicle sensor, a 3 is the vehicle positioning data information, η 1 , η 2 and η 3 is the weight factor of the chaotic mapping, and the vehicle data is fused to obtain the fused vehicle data information;

[0094] U32. Based on the data information of the fused vehicle, a polynomial feature extraction function F of the vehicle is established.

[0095]

[0096] Among them, b is the data information of the vehicle after fusion processing, λ 0 , 1 , 2 and λ 3 Extracting factors for the polynomial features of the vehicle;

[0097] U33. Based on the polynomial feature extraction function F of the vehicle, the geometric information of the road, the position of the traffic sign and the information of the obstacle distribution are extracted to obtain the geometric information of the road, the position of the traffic sign and the information of the obstacle distribution.

[0098] In this embodiment, the polynomial feature extraction factor λ of the vehicle 0 , 1 , 2 and λ 3 The constraints are:

[0099]

[0100] Example 2: Based on the method for generating a passenger car point cloud map based on a simulation engine in Example 1, the present invention is further illustrated and described below.

[0101] like Figure 1 As shown, a method for generating a point cloud map of a passenger car based on a simulation engine, the method comprising:

[0102] U1. Based on the data information of high-precision maps, the dynamic model of the vehicle and the simulation engine, a multi-modal integrated optimization algorithm based on refractive reverse learning mechanism and adaptive control factor is used for integration and optimization to build a high-precision simulation environment for the vehicle;

[0103] U2. Based on the high-precision simulation environment of the vehicle, the data information of the parameters of the vehicle controller and the data information of the parameters of the vehicle positioning controller are input to simulate the vehicle, and the data information of the vehicle's motion state, the data information of the sensor and the positioning data information are output;

[0104] U3. Based on the data information of the motion state of the vehicle, the data information of the sensor and the positioning data information, a polynomial feature extraction algorithm based on chaotic mapping is used to extract the geometric information of the road, the position of the traffic sign and the information of the obstacle distribution, and obtain the geometric information of the road, the position of the traffic sign and the information of the obstacle distribution;

[0105] U4. Based on the geometric information of the road, the location of traffic signs and the information of obstacle distribution, a point cloud map generation model of the road is constructed, and the point cloud map of the road is constructed to obtain data information of the point cloud map of the road.

[0106] In this embodiment, if Figure 4 As shown, in step U4, the point cloud map generation model of the road is constructed, and the point cloud map of the road is constructed including:

[0107] U41. Based on the geometric information of the road, the location of traffic signs and the information of obstacle distribution, data cleaning, denoising and calibration are performed to obtain the processed geometric information of the road, the location of traffic signs and the information of obstacle distribution;

[0108] U42. Based on the processed road geometry information, traffic sign location and obstacle distribution information, establish a road point cloud map generation function G,

[0109]

[0110] Among them, c 1 is the geometric information of the processed road, c 2 is the position of the traffic sign after processing, c 3 is the obstacle distribution information after processing, μ 1 , μ 2 and μ 3 Determinants of point cloud map generation;

[0111] U43. Based on the point cloud map generation function G of the road, the point cloud map of the road is constructed to obtain data information of the point cloud map of the road.

[0112] In this embodiment, the determining factor μ for generating the point cloud map is 1 , μ 2 and μ 3 The constraint function L is,

[0113]

[0114] Among them, the value range of the constraint function L is (3,5).

[0115] like Figure 5 As shown, simulation engine: VTD is selected as the simulation engine because it has real-time, image-based point cloud processing capabilities and is suitable for creating real-time, interactive point cloud maps. VTD provides an image input interface for reading point cloud data and converting it into a suitable format.

[0116] like Figure 6 As shown in the figure, the sensor simulation module: simulates the transmission and reception of high-frequency positioning signals (such as IMU) to ensure that the positioning accuracy reaches the centimeter level. Simulates the point cloud data output of the Lidar sensor, including point cloud filtering, edge detection, straight line detection and other processing algorithms.

[0117] Environmental simulation module: Build a high-precision simulation environment, including road elements such as roads, buildings, and traffic signs, as well as traffic elements such as people, cars, and trees. Use deep learning technology to generate high-simulation, high-resolution 3D point cloud data to meet the needs of the autonomous driving vehicle system for point cloud data.

[0118] like Figure 7 or Figure 8 As shown in the figure, the data processing and map generation module: receives the point cloud data output by the sensor simulation module, performs preprocessing and feature extraction. It uses the automatic detection algorithm to extract road and environmental elements, and matches and fuses environmental elements with road elements. It generates a high-precision point cloud map and adds various geographic information elements, such as streets, buildings, annotations, etc.

[0119] Verification and release module: Perform manual sampling inspection on the generated point cloud map to make up for the errors in the automated processing process. Use the cloud platform to update and release high-precision maps to ensure the real-time and accuracy of map data.

[0120] Implementation steps

[0121] 1. Preparation stage:

[0122] Determine the simulation engine and required modules, such as VTD, sensor simulation module, environmental simulation module, etc.

[0123] Collect and process real-world road environment data to build a high-precision simulation environment.

[0124] 2. Simulation environment construction:

[0125] Use VTD's image input interface to read road environment data and build a simulation environment. Add traffic elements and road elements to the simulation environment to ensure the authenticity and integrity of the simulation environment.

[0126] 3. Sensor simulation and data acquisition:

[0127] Configure the sensor simulation module to simulate the high-frequency positioning signal and point cloud data output of the Lidar sensor.

[0128] Collect point cloud data generated in the simulation environment and perform preprocessing and feature extraction.

[0129] 4. Point cloud map generation:

[0130] The extracted features are matched and fused with road elements using the data processing and map generation modules.

[0131] Generate high-precision point cloud maps and add necessary geographic information elements.

[0132] 5. Verification and release:

[0133] Manual sampling inspection is performed on the generated point cloud map to ensure the accuracy and completeness of the map. The high-precision map is updated and published using the cloud platform for use by the autonomous driving vehicle system.

[0134] Key technical points and optimization

[0135] 1. Construction of high-precision simulation environment:

[0136] Deep learning technology is used to generate high-simulation and high-resolution 3D point cloud data to improve the realism and accuracy of the simulation environment.

[0137] The simulation environment is continuously optimized and adjusted to adapt to different road environments and traffic conditions.

[0138] 2. Optimization of sensor simulation and data acquisition:

[0139] Select high-precision sensor simulation modules to ensure positioning accuracy and point cloud data accuracy. Preprocess and extract features of collected point cloud data to improve the efficiency and accuracy of map generation.

[0140] 3. Optimization of point cloud map generation:

[0141] Utilize advanced automatic detection algorithms and matching fusion technology to improve the automation and accuracy of map generation.

[0142] Manual sampling inspection is performed on the generated point cloud map to make up for the errors and deficiencies in the automated processing.

[0143] Optimization of verification and release:

[0144] Establish a complete verification mechanism to ensure that the generated point cloud map meets the usage requirements of the autonomous driving vehicle system.

[0145] Use the cloud platform to achieve real-time update and release of high-precision maps, and improve the availability and real-time performance of map data.

[0146] In this embodiment, the present invention provides a passenger car point cloud map generation system based on a simulation engine, including a computer device, which is programmed or configured to execute any one of the steps of the passenger car point cloud map generation method based on a simulation engine.

[0147] In this embodiment, the present invention provides a computer-readable storage medium, on which is stored a computer program programmed or configured to execute any one of the methods for generating a passenger car point cloud map based on a simulation engine.

[0148] Any reference to memory, storage, database or other media used in the embodiments provided in the present application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM) and memory bus dynamic RAM (RDRAM).

[0149] In summary, the present invention not only improves the accuracy and compatibility of the simulation system, can simulate high-frequency positioning signals and Lidar point cloud data, and meet the needs of point cloud map generation, but also provides a new and efficient method for the generation of passenger car point cloud maps, reducing production costs and cycles.

[0150] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A method for generating a point cloud map of a passenger car based on a simulation engine, characterized in that: The method comprises: U1. Based on the data information of high-precision maps, the dynamic model of the vehicle and the simulation engine, a multi-modal integrated optimization algorithm based on refractive reverse learning mechanism and adaptive control factor is used for integration and optimization to build a high-precision simulation environment for the vehicle; U2. Based on the high-precision simulation environment of the vehicle, the data information of the parameters of the vehicle controller and the data information of the parameters of the vehicle positioning controller are input to simulate the vehicle, and the data information of the vehicle's motion state, the data information of the sensor and the positioning data information are output; U3. Based on the data information of the motion state of the vehicle, the data information of the sensor and the positioning data information, a polynomial feature extraction algorithm based on chaotic mapping is used to extract the geometric information of the road, the position of the traffic sign and the information of the obstacle distribution, and obtain the geometric information of the road, the position of the traffic sign and the information of the obstacle distribution; U4. Based on the geometric information of the road, the location of traffic signs and the information of obstacle distribution, a point cloud map generation model of the road is constructed, and the point cloud map of the road is constructed to obtain data information of the point cloud map of the road.

2. The method for generating a point cloud map of a passenger car based on a simulation engine according to claim 1, characterized in that: In step U1, the integration and optimization using a multi-modal integrated optimization algorithm based on a refraction reverse learning mechanism and an adaptive control factor includes: U1. Based on the data information of the high-precision map and the dynamic model of the vehicle, an interaction function Q based on the refraction reverse learning mechanism is established. Among them, x is the data information of the high-precision map, y is the dynamic model of the vehicle, α1, α2 and α3 are interactive learning factors, and f is the refraction reverse learning mechanism function, which characterizes the interactive information between the vehicle and the high-precision map and obtains the interactive data information between the vehicle and the high-precision map; U12. Based on the data information of the interaction between the vehicle and the high-precision map and the simulation engine, a multimodal integrated function W based on an adaptive control factor is established. Among them, r is the data information of the interaction between the vehicle and the high-precision map, h is the simulation engine, β1, β2 and β3 are adaptive control factors; U13. Based on the multi-modal integrated function W based on the adaptive control factor, integration and optimization are performed to build a high-precision simulation environment for the vehicle.

3. The method for generating a point cloud map of a passenger car based on a simulation engine according to claim 2, characterized in that: The constraint function g of the adaptive control factors β1, β2 and β3 is, Among them, the value range of the constraint function g is (0,1).

4. The method for generating a point cloud map of a passenger car based on a simulation engine according to claim 2, characterized in that: The constraints of the interactive learning factors α1, α2 and α3 are: The refraction reverse learning mechanism function f is, Among them, y is the dynamic model of the vehicle.

5. The method for generating a point cloud map of a passenger car based on a simulation engine according to claim 1, characterized in that: The simulation engine includes an IMU data simulation model and a laser point cloud data simulation model. The IMU data simulation model establishes an IMU data simulation function R according to the data information of the parameters of the vehicle controller and the data information of the parameters of the vehicle positioning controller. Among them, z1 is the data information of the parameters of the vehicle controller, z2 is the data information of the parameters of the vehicle positioning controller, δ1, δ2 and δ3 are the error parameter factors of IMU data simulation, and the IMU positioning data of the vehicle is simulated; the laser point cloud data simulation model is to establish a laser point cloud data simulation function P according to the data information of the parameters of the vehicle controller and the data information of the parameters of the vehicle positioning controller, Among them, z1 is the data information of the parameters of the vehicle controller, z2 is the data information of the parameters of the vehicle positioning controller, γ1, γ2 and γ3 are the error factors of the laser point cloud data simulation, and the laser point cloud data of the vehicle is simulated.

6. The method for generating a point cloud map of a passenger car based on a simulation engine according to claim 1, characterized in that: In step U3, the extraction of road geometry information, traffic sign locations, and obstacle distribution information using a polynomial feature extraction algorithm based on chaotic mapping includes: U31. Based on the data information of the vehicle's motion state, the sensor data information and the positioning data information, establish the vehicle's chaotic mapping function S, Among them, a1 is the data information of the vehicle's motion state, a2 is the data information of the vehicle's sensor, a3 is the data information of the vehicle's positioning, η1, η2 and η3 are the weight factors of the chaotic mapping, and the vehicle data is fused to obtain the fused vehicle data information; U32. Based on the data information of the fused vehicle, a polynomial feature extraction function F of the vehicle is established. Among them, b is the data information of the vehicle after fusion processing, λ0, λ1, λ2 and λ3 are the polynomial feature extraction factors of the vehicle; U33. Based on the polynomial feature extraction function F of the vehicle, the geometric information of the road, the position of the traffic sign and the information of the obstacle distribution are extracted to obtain the geometric information of the road, the position of the traffic sign and the information of the obstacle distribution.

7. The method for generating a point cloud map of a passenger car based on a simulation engine according to claim 6, characterized in that: The constraints of the polynomial feature extraction factors λ0, λ1, λ2 and λ3 of the vehicle are:

8. The method for generating a point cloud map of a passenger car based on a simulation engine according to claim 1, characterized in that: In step U4, the step of constructing a point cloud map generation model for a road includes: U41. Based on the geometric information of the road, the location of traffic signs and the information of obstacle distribution, data cleaning, denoising and calibration are performed to obtain the processed geometric information of the road, the location of traffic signs and the information of obstacle distribution; U42. Based on the processed road geometry information, traffic sign location and obstacle distribution information, establish a road point cloud map generation function G, Among them, c1 is the processed road geometry information, c2 is the processed traffic sign position, c3 is the processed obstacle distribution information, μ1, μ2 and μ3 are the determining factors for point cloud map generation; U43. Based on the point cloud map generation function G of the road, the point cloud map of the road is constructed to obtain data information of the point cloud map of the road.

9. The method for generating a point cloud map of a passenger car based on a simulation engine according to claim 8, characterized in that: The constraint function L of the determining factors μ1, μ2 and μ3 for generating the point cloud map is, Among them, the value range of the constraint function L is (3,5).

10. A passenger car point cloud map generation system based on a simulation engine, comprising a computer device, characterized in that: The computer device is programmed or configured to execute the steps of the method for generating a passenger car point cloud map based on a simulation engine as described in any one of claims 1 to 9.

Images