CN115235452A - Intelligent parking positioning system and method based on UWB/IMU and visual information fusion - Google Patents

Abstract

The invention relates to an intelligent parking positioning system and method based on the fusion of UWB/IMU and visual information, which is used for vehicle positioning in a parking lot, and includes a central control module, a UWB/IMU module, a camera unit, a signal transmission and processing module, a computing Together with the positioning display module, the vehicle trajectory analysis module and the data fusion module, the central control module includes a vehicle information quantification unit, a camera mechanism threshold unit, an environmental error neural network learning model, a path guidance unit and a parking space guidance unit. Compared with the prior art, the present invention is assisted by the parking lot terminal equipment in the intelligent parking process to realize dual-model fusion positioning, design an environmental error neural network learning model to eliminate errors and improve accuracy, and determine the camera deflection angle according to the number of vehicles, so that the The parking lot camera mechanism dynamically monitors every moving vehicle in the parking lot, and the vehicle can track its real-time high-precision position in an unfamiliar parking environment, and realize the intelligent parking process through the cooperation between the parking lot and the vehicle.

Description

Intelligent parking positioning system and method based on UWB/IMU and visual information fusion

technical field

The invention relates to the technical field of vehicle intelligent parking positioning, in particular to an intelligent parking positioning system and method based on UWB/IMU and visual information fusion.

Background technique

With the continuous development of science and technology and the continuous increase of car ownership, the intelligence of cars has also been further developed. Among them, various car manufacturers have also carried out corresponding intelligent upgrades for the parking function of vehicles. As intelligent parking is an important part of the "last mile" in autonomous driving, commercialization will be prioritized, and intelligent parking systems have become an important direction for the research and development of various car companies.

At present, most of the market solutions mainly rely on pure vehicle-end implementation, such as building a three-dimensional map through vehicle lidar to perceive the surrounding environment and vehicle vision to scan the environment to obtain information to achieve the purpose of parking. However, due to the lack of scope and high price of lidar, it cannot be promoted in the market, and pure vision is obviously disturbed by the environment. At the same time, it needs to learn to park in an unfamiliar parking lot. The commercialization of parking cannot be well promoted and realized.

At the same time, due to the continuous increase in the number of cars and the increasing gap of parking spaces, the dense areas of major cities are faced with the problem of parking difficulties. Drivers cannot achieve "three-minute happy parking" during peak hours. Even smart vehicles with parking need to The driver gives the corresponding operation.

In summary, the existing parking solutions need to be improved to overcome the shortcomings of pure vehicle-side intelligent parking.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide an intelligent parking positioning system and method based on UWB/IMU and visual information fusion in order to overcome the above-mentioned defects in the prior art.

The object of the present invention can be realized through the following technical solutions:

An intelligent parking positioning system based on UWB/IMU and visual information fusion, used for vehicle positioning in parking lots, including a central control module, a UWB/IMU module, a camera unit, a signal transmission and processing module, a calculation and positioning display module, A vehicle trajectory analysis module and a data fusion module, the central control module includes a vehicle information quantification unit, a camera mechanism threshold unit, an environmental error neural network learning model, a path guidance unit and a parking space guidance unit;

The camera unit includes a plurality of cameras, which are used to obtain the current image information of the parking lot and track the target vehicle and the scene where the vehicle is located in real time, obtain the motion equation, and determine the vehicle position information of the target vehicle, and the current image information of the parking lot includes the vehicle image. information, image information of obstacles, and image information of surrounding parking spaces and lane lines;

The vehicle information quantification unit is used to quantify the image information collected by the camera unit into vehicle information, determine the number of vehicles in the current scene collected by the camera unit, and calibrate the target vehicle;

The camera mechanism threshold unit is used to determine the deflection angle threshold of the camera unit according to the number of vehicles in the current scene and control the camera unit to perform;

The vehicle trajectory analysis module is used to obtain the trajectory information of the target vehicle according to the vehicle image information at successive moments, and transmit it to the central control module for learning by the environmental error neural network learning model;

The UWB/IMU module is used to obtain the distance between the target vehicle and the UWB base station and the motion information of the vehicle, so as to obtain the virtual coordinate information and inertial direction of the target vehicle in the whole area of the parking lot;

In the environmental error neural network learning model, a convolutional neural network is used to establish an environmental error perception deep learning model, and an error factor of the positioning deviation caused by environmental factors is extracted to help the UWB/IMU module and the camera unit to correct the positioning accuracy;

The signal transmission and processing module is used to transmit the data of the UWB/IMU module and the camera unit to the central control module;

The calculation and positioning display module is used for coordinate calculation and visual position visual tracking display according to the data of the UWB/IMU module and the camera unit;

The data fusion module is used to fuse the virtual coordinate information of the UWB/IMU module and the vehicle position information of the target vehicle of the camera unit to obtain real-time precise position information, and transmit the fusion result to the central control module;

The path guidance unit is used to filter out the lane information with the least passing vehicles according to the current image information of the parking lot;

The parking space guidance unit is used for filtering out empty parking spaces according to the current image information of the parking lot.

Preferably, in the environmental error neural network learning model, a convolutional neural network is used to establish a deep learning model for environmental error perception, to extract the error factors of the positioning deviation caused by environmental factors, and to combine abstraction layer by layer to generate high-level features to help UWB/IMU The module corrects the positioning accuracy; the method of extracting the error factor of the positioning deviation caused by the environmental factor includes:

The nth fixed UWB base station coordinates U _n =(x _n , y _n , z _n ) are known coordinates; the position of the vehicle to be positioned at time t is denoted as N _t =(x _t , y _t , z _t ); The distance from the UWB base station to the target vehicle at time t is:

为此时误差因子；in

error factor for this time;

用以环境误差神经网络模型学习；其中

是高层特征量；

为权和系数，

其中v为目标车辆行驶速度；T_i+1、T_i对应目标车辆行进过程中某一时刻及后一时间帧记录的时间，(T_i+1-T_i)为行进的时间差；θ_i为此时刻车辆轮转角度。Substitute the error factors at different times into

used for environmental error neural network model learning; where

is the high-level feature quantity;

is the weight and coefficient,

Among them, v is the speed of the target vehicle; T _i+1 and T _i correspond to the time recorded at a certain moment and the next time frame during the travel of the target vehicle, (T _i+1 -T _i ) is the travel time difference; θ _i is The rotation angle of the vehicle at this moment.

Preferably, the vehicle information quantification unit is used to quantify the image information collected by the camera unit into vehicle information, determine the number of vehicles in the current scene collected by the camera unit, and calibrate the target vehicle, specifically:

The vehicle information quantification unit obtains the vehicle image information, quantifies the image information into the model, color and license plate number of the vehicle corresponding to the pixel point, generates a unique string code for storage in sequence, and generates a vehicle digital ID for each vehicle; the vehicle information quantification unit Obtain the body image of the current time frame of the target vehicle, perform local image processing on the body image, obtain discrete pixels corresponding to the body, color and license plate, and then convert them into discrete quantitative values to generate a unique vehicle digital ID within the corresponding time, as the target Vehicles are identified.

Preferably, the camera mechanism threshold unit is used to determine the deflection angle threshold of the camera unit according to the number of vehicles in the current scene and control the camera unit to perform, specifically:

The azimuth corresponding to the ith high-precision camera and the parking space coordinates (x, y, z) is (α _i , β _i , γ _i ), and (x, y, z) corresponds to the ith high-precision camera The spatial coordinate position in the parking lot, and the number of target vehicles within the viewing angle is collected according to this azimuth angle as N _k , and the state matrix equation is constructed:

Among them, the analytical computing power of the camera is R _χ , and ξ is the threshold value set by the camera mechanism. When ξ≤N _k , the camera mechanism threshold unit sends a deflection command to the camera to realize the camera angle deflection.

Preferably, the UWB/IMU module is used to obtain the distance between the target vehicle and the UWB base station and the motion information of the vehicle, so as to obtain the virtual coordinate information and inertial direction of the target vehicle in the whole area of the parking lot, specifically:

The UWB/IMU module obtains the distance between the target vehicle and the base station: obtains the spatial coordinates of each UWB base station in the parking lot, obtains the time transmitted by the pulse signal between each UWB base station and the target vehicle, and calculates the distance between each base station and the target vehicle. distance, so that the virtual coordinate information of the target vehicle in the parking lot can be calculated:

Among them, m and n are used to identify different base stations, lm _{, n} represent the distance between UWB base stations m and n, t is the pulse transmission time; c is the speed of light; (x, y, z) is the target vehicle in the parking lot Inner virtual coordinates;

The UWB/IMU module obtains the motion information of the vehicle: the accelerometer data E(ε) and the gyroscope data E(σ) are obtained through the IMU inertial module, so as to obtain the inertial travel direction of the target vehicle.

Preferably, the data fusion module is used to fuse the virtual coordinate information of the UWB/IMU module and the vehicle position information of the target vehicle of the camera unit, and transmit the fusion result to the central control module, specifically:

Establish the fusion target positioning optimization function:

G _i =f(x _i-1 , _ui , _wi )

H _i,j =h(y _j , _xi ,vi _,j )

为UWB定位处理后坐标数据，T为观测目标车辆时间帧，w_i为车辆响应速率，G_i为摄像机单元跟踪目标车辆得到的运动方程，H_i,j为车辆轨迹分析模块确定的轨迹预测方程，u_i、v_i,j为观测噪声，x_i为目标车辆位置，y_j为车位的坐标；Among them, E(ε) and E(σ) are the accelerometer data and gyroscope data measured by the MU inertial module,

is the coordinate data after UWB positioning processing, T is the time frame of the observed target vehicle, _wi is the response rate of the vehicle, G _i is the motion equation obtained by the camera unit tracking the target vehicle, H _i,j is the trajectory prediction equation determined by the vehicle trajectory analysis module , u _i , vi _,j are the observation noise, _xi is the target vehicle position, y _j is the coordinates of the parking space;

When the target positioning optimization function solves the minimum point, it is the real-time precise position information of the final optimized vehicle.

An intelligent parking positioning method based on UWB/IMU and visual information fusion, based on the above-mentioned intelligent parking positioning system based on UWB/IMU and visual information fusion, comprising the following steps:

S1, obtain the current image information of the parking lot through the camera unit and track the target vehicle and the scene where the vehicle is located in real time, obtain the motion equation, and determine the vehicle position information of the target vehicle, where the current image information of the parking lot includes vehicle image information and obstacle image information. and surrounding parking spaces and lane line image information;

S2, quantifying the image information collected by the camera unit into vehicle information through the vehicle information quantification unit, determining the number of vehicles in the current scene collected by the camera unit, and calibrating the target vehicle;

S3, according to the number of vehicles in the current scene, determine the deflection angle threshold of the camera unit through the camera mechanism threshold unit and control the camera unit to execute;

S4. The UWB/IMU module obtains the distance between the target vehicle and the UWB base station and the motion information of the vehicle, so as to obtain the virtual coordinate information and inertial direction of the target vehicle in the whole area of the parking lot;

S5. The vehicle trajectory analysis module obtains the trajectory information of the target vehicle according to the vehicle image information at successive moments, and transmits it to the central control module for learning by the environmental error neural network learning model;

S6. In the environmental error neural network learning model, a convolutional neural network is used to establish an environmental error perception deep learning model, and the error factor of the positioning deviation caused by environmental factors is extracted to help the UWB/IMU module and the camera unit to correct the positioning accuracy;

S7. The signal transmission and processing module transmits the data of the UWB/IMU module and the camera unit to the central control module;

S8, the calculation and positioning display module performs coordinate calculation and visual position visual tracking display according to the data of the UWB/IMU module and the camera unit;

S9, the data fusion module fuses the virtual coordinate information of the UWB/IMU module and the vehicle position information of the target vehicle of the camera unit to obtain the real-time precise position information of the vehicle, and transmits the fusion result to the central control module;

S10, the path guidance unit filters out the lane information with the least passing vehicles according to the current image information of the parking lot, the parking space guidance unit filters out the empty parking spaces according to the current image information of the parking lot, and the central control module according to the real-time precise position information of the vehicle and the lane information and empty parking spaces The coordinates guide the target vehicle to park.

Preferably, in step S6, the method of extracting the error factor for the positioning deviation caused by the environmental factor includes:

The nth fixed UWB base station coordinates U _n =(x _n , y _n , z _n ) are known coordinates; the position of the vehicle to be positioned at time t is denoted as N _t =(x _t , y _t , z _t ); The distance from the UWB base station to the target vehicle at time t is:

为此时误差因子：in

For this time the error factor:

用以环境误差神经网络模型学习；其中

是高层特征量；

为权和系数，

其中v为目标车辆行驶速度；T_i+1、T_i对应目标车辆行进过程中某一时刻及后一时间帧记录的时间，(T_i+1-T_i)为行进的时间差；θ_i为此时刻车辆轮转角度。Substitute the error factors at different times into

used for environmental error neural network model learning; where

is the high-level feature quantity;

is the weight and coefficient,

Among them, v is the speed of the target vehicle; T _i+1 and T _i correspond to the time recorded at a certain moment and the next time frame during the travel of the target vehicle, (T _i+1 -T _i ) is the travel time difference; θ _i is The rotation angle of the vehicle at this moment.

Preferably, step S2 is specifically:

The vehicle information quantification unit obtains the vehicle image information, quantifies the image information into the model, color and license plate number of the vehicle corresponding to the pixel point, generates a unique string code for storage in sequence, and generates a vehicle digital ID for each vehicle; the vehicle information quantification unit Obtain the body image of the current time frame of the target vehicle, perform local image processing on the body image, obtain discrete pixels corresponding to the body, color and license plate, and then convert them into discrete quantitative values to generate a unique vehicle digital ID within the corresponding time, as the target Vehicles are identified.

Preferably, step S3 is specifically:

The azimuth corresponding to the ith high-precision camera and the parking space coordinates (x, y, z) is (α _i , β _i , γ _i ), and (x, y, z) corresponds to the ith high-precision camera The spatial coordinate position in the parking lot, and the number of target vehicles within the viewing angle is collected according to this azimuth angle as N _k , and the state matrix equation is constructed:

Among them, the analytical computing power of the camera is R _χ , and ξ is the threshold value set by the camera mechanism. When ξ≤N _k , the camera mechanism threshold unit sends a deflection command to the camera to realize the camera angle deflection.

Preferably, step S4 is specifically:

The UWB/IMU module obtains the distance between the target vehicle and the base station: obtains the spatial coordinates of each UWB base station in the parking lot, obtains the time transmitted by the pulse signal between each UWB base station and the target vehicle, and calculates the distance between each base station and the target vehicle. distance, so that the virtual coordinate information of the target vehicle in the parking lot can be calculated:

Among them, m and n are used to identify different base stations, lm _{, n} represent the distance between UWB base stations m and n, t is the pulse transmission time; c is the speed of light; (x, y, z) is the target vehicle in the parking lot Inner virtual coordinates;

The UWB/IMU module obtains the motion information of the vehicle: the accelerometer data E(ε) and the gyroscope data E(σ) are obtained through the IMU inertial module, so as to obtain the inertial travel direction of the target vehicle.

Preferably, step S9 is specifically:

Establish the fusion target positioning optimization function:

G _i =f(x _i-1 , _ui , _wi )

H _i,j =h(y _j , _xi ,vi _,j )

为UWB定位处理后坐标数据，T为观测目标车辆时间帧，w_i为车辆响应速率，G_i为摄像机单元跟踪目标车辆得到的运动方程，H_i,j为车辆轨迹分析模块确定的轨迹预测方程，u_i、v_i,j为观测噪声，x_i为目标车辆位置，y_j为车位的坐标；Among them, E(ε) and E(σ) are the accelerometer data and gyroscope data measured by the MU inertial module,

is the coordinate data after UWB positioning processing, T is the time frame of the observed target vehicle, _wi is the response rate of the vehicle, G _i is the motion equation obtained by the camera unit tracking the target vehicle, H _i,j is the trajectory prediction equation determined by the vehicle trajectory analysis module , u _i , vi _,j are the observation noise, _xi is the target vehicle position, y _j is the coordinates of the parking space;

When the target positioning optimization function solves the minimum point, it is the real-time precise position information of the final optimized vehicle.

Compared with the prior art, the present invention has the following beneficial effects:

(1) In the process of intelligent parking, with the assistance of the parking lot terminal equipment, the dual-model fusion positioning is realized, and the real-time high-precision position tracking of the vehicle in an unfamiliar parking environment can be solved, and the intelligent parking process can be realized through the cooperation between the parking lot and the vehicle.

(2) The environmental error neural network learning model is designed, and a convolutional neural network is used to establish a deep learning model for environmental error perception, extracting the error factors of the positioning deviation caused by environmental factors, and combining abstraction layer by layer to generate high-level features, which can remove the error factors and improve the Target vehicle positioning accuracy.

(3) The camera mechanism threshold unit determines the camera deflection angle according to the number of vehicles in the current scene. The purpose is to enable the parking lot camera mechanism to dynamically monitor every moving vehicle in the parking lot, and each camera will not monitor too many vehicles, so as not to calculate Insufficient power to track the target vehicle.

Description of drawings

1 is a schematic structural diagram of an intelligent parking positioning system;

Fig. 2 is the structural representation of the central control module;

3 is a flowchart of an intelligent parking positioning method;

4 is a schematic diagram of a usage scenario of an intelligent parking positioning system;

5 is a flowchart of an intelligent parking method based on UWB/IMU and visual information fusion in an embodiment of the present invention;

6 is a flowchart of a method for removing environmental error factor fusion positioning in an embodiment of the present invention;

FIG. 7 is a flowchart of a method for fusion sending of positioning information in an embodiment of the present invention;

Reference numerals: 1. Central control module, 2. UWB/IMU module, 3. Camera unit, 4. Signal transmission and processing module, 5. Calculation and positioning display module, 6. Vehicle trajectory analysis module, 7. Data fusion module , 11, vehicle information quantification unit, 12, camera mechanism threshold unit, 13, environment error neural network learning model, 14, path guidance unit, 15, parking space guidance unit.

Detailed ways

In order to further understand the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. The embodiment is implemented on the premise of the technical solution of the present invention, and provides a detailed implementation manner and a specific operation process. It should be understood that these descriptions are intended to further illustrate the features and advantages of the present invention, rather than to limit the claims of the present invention. The description in this section is only for a few typical embodiments, and the present invention is not limited to the scope of the description of the embodiments. It is also within the scope of the description and protection of the present invention to replace some technical features in the embodiments with the same or similar prior art means.

Reference herein to "one embodiment" or "an embodiment" refers to a particular feature, structure, or characteristic that may be included in at least one implementation of the present invention. In the description of the present invention, it is to be understood that the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units is not limited to the listed steps or units, but optionally also includes unlisted steps or units, or optionally also includes For other steps or units inherent to these processes, methods, products or devices.

Example 1:

An intelligent parking positioning system based on UWB/IMU and visual information fusion, used for vehicle positioning in parking lots, as shown in Figure 1, including a central control module, UWB/IMU module, camera unit, signal transmission and processing module, Calculation and positioning display module, vehicle trajectory analysis module and data fusion module, as shown in Figure 2, the central control module includes a vehicle information quantification unit, a camera mechanism threshold unit, an environmental error neural network learning model, a path guidance unit and a parking space guidance unit. In this embodiment, the application scenario is shown in Figure 4. There are multiple UWB base stations, which are installed in the parking lot, UWB transmitters are installed on the target vehicle, and the camera unit includes multiple cameras, which are installed in the parking lot. The signal transmission and processing unit realizes For the data transmission between the target vehicle and the parking lot, the IMU inertial module is installed on the target vehicle.

In this embodiment, the present invention mainly realizes parking positioning by merging two parts of UWB/IMU positioning and vision-assisted positioning at the parking lot end. On the one hand, the present invention uses the UWB/IMU module to obtain the distance between the target vehicle and the base station, so as to obtain the virtual coordinate information and inertial travel direction of the target vehicle in the whole area of the parking lot. On the other hand, the high-precision camera of the present invention obtains the parking lot. The image information of the target vehicle, obstacles and surrounding parking space lines is used to collect the image information of the body and the environment in the parking lot environment where the target vehicle is located at the current time frame, and the coordinate position information of the empty parking space, etc. Coordinate information and vehicle location information tracked by high-precision cameras, and determine the coordinates of empty parking spaces and lanes with fewer passing vehicles, and cooperate with target vehicles to achieve path planning for designated empty parking spaces, thereby realizing intelligent parking.

At the same time, an environmental error neural network learning model is established in combination with the environmental errors existing in different parking lots, and the errors are analyzed and eliminated to improve the accuracy of the positioning information collected by the module; the vehicle trajectory analysis module is used to obtain the trajectory information of the target vehicle, and the target vehicle is The position information at different times, and the analysis results are used for the learning of the environmental error neural network. In the environmental error neural network learning model, a convolutional neural network is used to establish a deep learning model for environmental error perception, to extract the error factors of the positioning deviation caused by environmental factors, and to abstractly generate high-level features layer by layer, which is used to help UWB/IMU modules and high-precision cameras. Correct positioning accuracy.

Moreover, for the images collected by the camera unit, the application has carried out information quantification processing, quantifying the collected image information into the model, color and license plate number of the vehicle corresponding to the pixel point, and generating a unique string code in sequence for storage; on the one hand, the target can be calibrated On the other hand, it can be analyzed by the threshold setting unit of the camera mechanism, and the number of vehicles in the current high-precision camera acquisition scene can be judged according to the number of received coded numbers, so as to set the high-precision camera deflection angle threshold for different numbers of vehicles, thereby Realize the dynamic monitoring of every moving vehicle in the parking lot by the parking lot camera mechanism.

This application designs an intelligent parking positioning system based on the fusion of UWB/IMU and visual information, and performs corresponding intelligent upgrades in the parking lot, integrating UWB/IMU modules and visual information, which can overcome the shortcomings of pure vehicle-side intelligent parking. Promote the commercialization of smart parking as soon as possible.

Specifically, the work of each module unit in an intelligent parking positioning system based on UWB/IMU and visual information fusion is as follows:

(1) The camera unit includes a plurality of cameras, which are used to obtain the current image information of the parking lot and track the target vehicle and the scene where the vehicle is located in real time, and obtain the motion equation G _i =f(x _i-1 , u _i , w _i ), where u _i and _xi are the position of the target vehicle, and the vehicle position information of the target vehicle is determined. The current image information of the parking lot includes the image information of the vehicle, the image information of the obstacles, and the image information of the surrounding parking spaces and lane lines; the spatial coordinates of each camera in the parking lot It is known, and its parameters such as the deflection angle and the focal length of the shooting are also known. Therefore, only after completing the calibration and analyzing the images collected by the camera, the vehicles, obstacles, lane lines, parking spaces, etc. in the image can be determined. The spatial coordinates within the parking lot.

In this embodiment, the lane line elements, the parking space elements and the classification of obstacles are calibrated at the same time, and the lane line element calibration is performed according to the current forms of common white dotted lines, yellow solid lines and white left and right turn arrows in the parking lot. , the three forms are respectively collected by high-precision cameras and sent to the central control module of the target vehicle. The parking space elements are calibrated according to the three common types of parking spaces in the market: vertical, horizontal and inclined. These three forms are respectively collected by high-precision cameras and sent to the central control module of the target vehicle. The classification and calibration of obstacles is based on the current parking lot. Common obstacles mainly include vehicles, pets, pedestrians and traffic signs. These four forms are collected by high-precision cameras and sent to the central control module of the target vehicle to realize blind spot collision warning.

(2) The vehicle information quantification unit is used to quantify the image information collected by the camera unit into vehicle information, determine the number of vehicles in the current scene collected by the camera unit, and calibrate the target vehicle;

The vehicle information quantification unit obtains the body image information in the parking lot where the target vehicle is currently located, quantifies the image information into the model, color and license plate number of the vehicle corresponding to the pixel point, generates a unique string code in sequence, and stores it for each vehicle. Generate the vehicle digital ID; the vehicle information quantification unit receives the current time frame body image of the target vehicle collected by the high-precision camera, performs local image processing on the body image, obtains the discrete pixels corresponding to the body, color and license plate, and then converts them into discrete quantities. value, generate a unique vehicle digital ID within the corresponding time, and perform identity calibration for the target vehicle.

The target vehicle is calibrated by the vehicle information quantification unit, and the real-time tracking of the target vehicle can be realized. The camera unit compares the data by transforming the encoded data of each frame before and after the image, and judges whether the data is consistent. If it is consistent, output the visual information data, otherwise search and collect again. After the image is encoded, the information is searched for matching data. Data can also be provided to the vehicle trajectory analysis unit.

(3) The camera mechanism threshold unit is used to determine the deflection angle threshold of the camera unit according to the number of vehicles in the current scene and control the camera unit to execute, in order to set the high-precision camera deflection angle threshold under different vehicle numbers;

The azimuth corresponding to the ith high-precision camera and the parking space coordinates (x, y, z) is (α _i , β _i , γ _i ), and (x, y, z) corresponds to the ith high-precision camera The spatial coordinate position in the parking lot, and the number of target vehicles within the viewing angle is collected according to this azimuth angle as N _k , and the state matrix equation is constructed:

Among them, the analytical computing power of the camera is R _χ , and ξ is the threshold value set by the camera mechanism. When ξ≤N _k , the camera mechanism threshold unit sends a deflection command to the camera to realize the camera angle deflection.

It can be seen that the deflection angle of the camera in this application is determined according to the number of vehicles in the current scene. The purpose is to enable the parking lot camera mechanism to dynamically monitor every moving vehicle in the parking lot, and each camera will not monitor too many vehicles to avoid Insufficient computing power to track the target vehicle.

(4) The vehicle trajectory analysis module is used to obtain the trajectory information of the target vehicle according to the vehicle image information at successive times, and transmit it to the central control module for learning by the environmental error neural network learning model;

The vehicle trajectory analysis module determines the trajectory analysis target vehicle and its characteristic information through a high-precision camera, and matches and locks the target vehicle position information according to the characteristic information and the vehicle information quantification unit. The vehicle travel direction is obtained by detecting the vehicle tire profile and the deflection angle, and the target vehicle position update information in different time frames is used to obtain the vehicle travel path, and the trajectory information of the target vehicle is obtained. The trajectory prediction equation is:

H _i,j =h(y _j , _xi ,vi _,j )

Among them, vi _,j is the observation noise; _xi is the target vehicle position; y _j is the parking space coordinate point.

By analyzing the vehicle trajectory and transmitting the analysis results to the central control module, it can be used for the learning of the environmental error neural network model, which can continuously optimize the entire system and improve the robustness and positioning accuracy of the entire intelligent parking system.

(5) The UWB/IMU module is used to obtain the distance between the target vehicle and the UWB base station and the motion information of the vehicle, so as to obtain the virtual coordinate information and inertial direction of the target vehicle in the whole area of the parking lot;

The UWB/IMU module obtains the distance between the target vehicle and the base station: obtains the spatial coordinates of each UWB base station in the parking lot, obtains the time transmitted by the pulse signal between each UWB base station and the target vehicle, and calculates the distance between each base station and the target vehicle. distance, so that the virtual coordinate information of the target vehicle in the parking lot can be calculated:

Among them, m and n are used to identify different base stations, lm _{, n} represent the distance between UWB base stations m and n, t is the pulse transmission time; c is the speed of light; (x, y, z) is the target vehicle in the parking lot Inner virtual coordinates;

In this embodiment, there are 4 UWB base stations in total, and their spatial coordinates in the parking lot are known, namely (x ₁ , y ₁ , z ₁ ), (x ₂ , y ₂ , z ₂ ), (x ₃ , y ₃ , z ₃ ), (x ₄ , y ₄ , z ₄ ), so by combining the above formulas, the virtual coordinates (x, y, z) of the target vehicle in the parking lot can be obtained.

The UWB/IMU module obtains the motion information of the vehicle: the accelerometer data E(ε) and the gyroscope data E(σ) are obtained through the IMU inertial module, so as to obtain the inertial travel direction of the target vehicle.

(6) In the environmental error neural network learning model, a convolutional neural network is used to establish a deep learning model for environmental error perception, to extract the error factors of the positioning deviation caused by environmental factors, and to abstractly generate high-level features layer by layer to help UWB/IMU modules and camera units. Correct the positioning accuracy; in the forward calculation of the network, in the convolution layer, there are multiple convolution kernels to perform convolution operations on the input at the same time, generating multiple feature maps, and the dimension of each feature map is reduced relative to the input dimension; In the sub-sampling layer, each feature map is pooled to obtain a corresponding map with a further reduced dimension. After being cross-stacked in turn, it reaches the network output through the fully connected layer for the entire intelligent parking system to actively learn and improve the robustness of the entire intelligent parking system. Robustness and positioning accuracy;

The nth fixed UWB base station coordinates U _n =(x _n , y _n , z _n ) are known coordinates; the position of the vehicle to be positioned at time t is denoted as N _t =(x _t , y _t , z _t ); The distance from the UWB base station to the target vehicle at time t is:

为此时误差因子；in

error factor for this time;

用以环境误差神经网络模型学习；其中

是高层特征量；

为权和系数，

其中v为目标车辆行驶速度；T_i+1、T_i对应目标车辆行进过程中某一时刻及后一时间帧记录的时间，(T_i+1-T_i)为行进的时间差；θ_i为此时刻车辆轮转角度。Substitute the error factors at different times into

used for environmental error neural network model learning; where

is the high-level feature quantity;

is the weight and coefficient,

Among them, v is the speed of the target vehicle; T _i+1 and T _i correspond to the time recorded at a certain moment and the next time frame during the travel of the target vehicle, (T _i+1 -T _i ) is the travel time difference; θ _i is The rotation angle of the vehicle at this moment.

It should be noted that the environmental error neural network learning model is used to correct the positioning accuracy. In practical applications, the camera unit can collect images and divide the scene into no environmental interference and environmental interference according to whether there is interference in the environment. Integrate the virtual coordinate information of the UWB/IMU module and the vehicle position information of the target vehicle of the camera unit to obtain real-time accurate position information to complete the pose estimation. For scenes with environmental interference, it is necessary to use the environmental error neural network learning model to help UWB/IMU The module and the camera unit correct the positioning accuracy, and then perform fusion to complete the pose estimation.

(7) The signal transmission and processing module is used to transmit the data of the UWB/IMU module and the camera unit to the central control module, mainly the current UWB positioning coordinates of the target vehicle and the quantized and encoded information of the surrounding environment image;

(8) The calculation and positioning display module is used for coordinate calculation and visual position visual tracking display according to the data of the UWB/IMU module and the camera unit. Specifically, it processes the coordinates and image information output by the UWB/IMU module and the high-precision camera, and performs UWB/IMU positioning coordinate calculation and visual position tracking display;

(9) The data fusion module is used to fuse the virtual coordinate information of the UWB/IMU module and the vehicle position information of the target vehicle of the camera unit to obtain real-time accurate position information, and transmit the fusion result to the central control module; the fusion process is:

Establish the fusion target positioning optimization function:

G _i =f(x _i-1 , _ui , _wi )

H _i,j =h(y _j , _xi ,vi _,j )

为UWB定位处理后坐标数据，T为观测目标车辆时间帧，w_i为车辆响应速率，G_i为摄像机单元跟踪目标车辆得到的运动方程，H_i,j为车辆轨迹分析模块确定的轨迹预测方程，u_i、v_i,j为观测噪声，x_i为目标车辆位置，y_j为车位的坐标，下标没有含义仅表示函数中代入数据进行计算；当目标定位优化函数求解最小点时为最终优化后车辆的实时精确位置信息，从而解决定位漂移及视觉定位偏差，提高定位精度。Among them, E(ε) and E(σ) are the accelerometer data and gyroscope data measured by the MU inertial module,

is the coordinate data after UWB positioning processing, T is the time frame of the observed target vehicle, _wi is the response rate of the vehicle, G _i is the motion equation obtained by the camera unit tracking the target vehicle, H _i,j is the trajectory prediction equation determined by the vehicle trajectory analysis module , u _i , vi _,j are the observation noise, _xi is the target vehicle position, y _j is the coordinates of the parking space, the subscript has no meaning and only means that the data is substituted into the function for calculation; when the target positioning optimization function solves the minimum point, it is the final The real-time accurate position information of the optimized vehicle can solve the positioning drift and visual positioning deviation and improve the positioning accuracy.

(10) The path guidance unit is used to filter out the lane information with the least passing vehicles according to the current image information of the parking lot; the parking space guidance unit is used to filter out the empty parking spaces according to the current image information of the parking lot; the central control module is based on the real-time precise position information of the vehicle and The lane information and the coordinates of the empty parking space guide the target vehicle to park.

The high-precision camera is used to obtain the image information of the vehicle's environment and the image information of the empty parking space in real time; the UWB/IMU module is used to obtain the distance information between the vehicle and the UWB base station of the parking lot; the signal transmission and processing module is used to receive and send the central control module The positioning signal sent; the calculation and positioning display module is used to process the position information of the vehicle in the simulated coordinates in real time; the vehicle trajectory analysis module is used to track the vehicle trajectory and upload it to the central control system to correct the positioning accuracy of the UWB/IMU module; the data fusion module It is used to solve positioning drift and visual positioning deviation and improve positioning accuracy. In the intelligent parking process, the invention realizes dual-model fusion positioning through the assistance of the parking lot terminal equipment, solves the real-time high-precision position tracking of the vehicle in the unfamiliar parking environment, and realizes the intelligent parking process through the cooperation between the parking lot equipment and the vehicle.

Example 2:

An intelligent parking positioning method based on UWB/IMU and visual information fusion, based on the intelligent parking positioning system described in Embodiment 1, the flowchart is shown in Figure 3, and Figures 5-7 can also be referred to to understand the For details, the present specification provides method operation steps such as embodiments or schematic flowcharts, but more or less operation steps may be included based on routine or non-creative work. The sequence of steps enumerated in the embodiments is only one of the execution sequences of many steps, and does not represent the only execution sequence. When an actual system or server product is executed, the methods shown in the embodiments or the accompanying drawings may be executed sequentially or in parallel (eg, in a parallel processor or multi-threaded processing environment), or the execution sequence of steps without timing constraints may be adjusted. Specifically, an intelligent parking positioning method based on UWB/IMU and visual information fusion includes the following steps:

S1. Obtain the current image information of the parking lot through the camera unit and track the target vehicle and the scene where the vehicle is located in real time, obtain the motion equation, and determine the vehicle position information of the target vehicle. The current image information of the parking lot includes vehicle image information, obstacle image information and surrounding Image information of parking spaces and lane lines;

S2, quantifying the image information collected by the camera unit into vehicle information through the vehicle information quantification unit, determining the number of vehicles in the current scene collected by the camera unit, and calibrating the target vehicle;

S3, according to the number of vehicles in the current scene, determine the deflection angle threshold of the camera unit through the camera mechanism threshold unit and control the camera unit to execute;

S4. The UWB/IMU module obtains the distance between the target vehicle and the UWB base station and the motion information of the vehicle, so as to obtain the virtual coordinate information and inertial direction of the target vehicle in the whole area of the parking lot;

S5. The vehicle trajectory analysis module obtains the trajectory information of the target vehicle according to the vehicle image information at successive moments, and transmits it to the central control module for learning by the environmental error neural network learning model;

S6. In the environmental error neural network learning model, a convolutional neural network is used to establish an environmental error perception deep learning model, and the error factor of the positioning deviation caused by environmental factors is extracted to help the UWB/IMU module and the camera unit to correct the positioning accuracy;

S7. The signal transmission and processing module transmits the data of the UWB/IMU module and the camera unit to the central control module;

S8, the calculation and positioning display module performs coordinate calculation and visual position visual tracking display according to the data of the UWB/IMU module and the camera unit;

S9, the data fusion module fuses the virtual coordinate information of the UWB/IMU module and the vehicle position information of the target vehicle of the camera unit to obtain the real-time precise position information of the vehicle, and transmits the fusion result to the central control module;

S10, the path guidance unit filters out the lane information with the least passing vehicles according to the current image information of the parking lot, the parking space guidance unit filters out the empty parking spaces according to the current image information of the parking lot, and the central control module according to the real-time precise position information of the vehicle and the lane information and empty parking spaces The coordinates guide the target vehicle to park.

In the above steps, the specific implementation details of each step are the same as those in Embodiment 1, and are not repeated here.

In a typical configuration of the present application, the terminal, the device serving the network, and the trusted party all include one or more processors (eg, a central processing unit (CPU)), an input/output interface, a network interface, and a memory .

Memory may include non-persistent memory in computer readable media, random access memory (RAM) and/or non-volatile memory, such as read only memory (ROM) or flash memory (flash). RAM). Memory is an example of a computer-readable medium.

Computer-readable media includes both persistent and non-permanent, removable and non-removable media, and storage of information may be implemented by any method or technology. Information may be computer readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (Phase-Change RAM, PRAM), static random access memory (Static Random Access Memory, SRAM), dynamic random access memory (Dynamic Random Access Memory, DRAM), Other types of random access memory (RAM), read only memory (ROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), flash memory or other memory technology, read-only Compact Disc Read-Only Memory (CD-ROM), Digital Versatile Disk (DVD) or other optical storage, magnetic cassette tape, magnetic tape disk storage or other magnetic storage device or any other non- A transmission medium that can be used to store information that can be accessed by a computing device. Computer-readable media, as defined herein, excludes non-transitory computer-readable media, such as modulated data signals and carrier waves.

It should be noted that the present application may be implemented in software and/or a combination of software and hardware, eg, an application specific integrated circuit (ASIC), a general purpose computer, or any other similar hardware device. In one embodiment, the software program of the present application may be executed by a processor to implement the steps or functions described above. Likewise, the software programs of the present application (including associated data structures) may be stored on a computer-readable recording medium, such as RAM memory, magnetic or optical drives or floppy disks, and the like. In addition, some steps or functions of the present application may be implemented in hardware, for example, as a circuit that cooperates with a processor to perform various steps or functions.

In addition, a part of the present application can be applied as a computer program product, such as computer program instructions, which when executed by a computer, through the operation of the computer, can invoke or provide methods and/or technical solutions according to the present application. The program instructions for invoking the methods of the present application may be stored in fixed or removable recording media, and/or transmitted via data streams in broadcast or other signal-bearing media, and/or stored in accordance with the in the working memory of the computer device on which the program instructions are executed. Here, an embodiment according to the present application includes an apparatus comprising a memory for storing computer program instructions and a processor for executing the program instructions, wherein, when the computer program instructions are executed by the processor, a trigger is The apparatus operates based on the aforementioned methods and/or technical solutions according to various embodiments of the present application.

The description and application of the present invention herein is illustrative, and is not intended to limit the scope of the present invention to the above-described embodiments. The effects or advantages involved in the embodiments may be interfered by various factors and may not be embodied in the embodiments, and the description of the effects or advantages is not intended to limit the embodiments. Variations and variations of the embodiments disclosed herein are possible, and alternative and equivalent various components of the embodiments are known to those of ordinary skill in the art. It should be apparent to those skilled in the art that the present invention may be implemented in other forms, structures, arrangements, proportions, and with other components, materials and components without departing from the spirit or essential characteristics of the invention. Other modifications and changes of the embodiments disclosed herein may be made without departing from the scope and spirit of the invention.

Claims (10)

1. An intelligent parking positioning system based on UWB/IMU and visual information fusion is characterized in that the system is used for vehicle positioning of a parking lot and comprises a central control module, a UWB/IMU module, a camera unit, a signal transmission and processing module, a calculation and positioning display module, a vehicle track analysis module and a data fusion module, wherein the central control module comprises a vehicle information quantification unit, a camera mechanism threshold value unit, an environmental error neural network learning model, a path guide unit and a parking space guide unit;

the camera unit comprises a plurality of cameras and is used for acquiring current image information of a parking lot, tracking a target vehicle and a scene where the vehicle is located in real time, obtaining a motion equation, and determining vehicle position information of the target vehicle, wherein the current image information of the parking lot comprises vehicle image information, obstacle image information and surrounding parking space and lane line image information;

the vehicle information quantization unit is used for quantizing the image information acquired by the camera unit into vehicle information, determining the number of vehicles in the current scene acquired by the camera unit and calibrating the target vehicle;

the camera mechanism threshold unit is used for determining a deflection angle threshold of the camera unit according to the number of vehicles in the current scene and controlling the camera unit to execute;

the vehicle track analysis module is used for acquiring track information of a target vehicle according to vehicle image information at continuous moments, and transmitting the track information to the central control module for learning of an environmental error neural network learning model;

the UWB/IMU module is used for acquiring the distance between a target vehicle and a UWB base station and the motion information of the vehicle so as to acquire the virtual coordinate information of the target vehicle in the whole parking lot domain and the inertial advancing direction;

in the environment error neural network learning model, a convolutional neural network is used for establishing an environment error perception deep learning model, extracting an error factor of positioning deviation caused by an environment factor, and helping a UWB/IMU module and a camera unit to correct positioning accuracy;

the signal transmission and processing module is used for transmitting data of the UWB/IMU module and the camera unit to the central control module;

the calculation and positioning display module is used for performing coordinate calculation and visual position visual tracking display according to the data of the UWB/IMU module and the camera unit;

the data fusion module is used for fusing virtual coordinate information of the UWB/IMU module and vehicle position information of a target vehicle of the camera unit to obtain real-time accurate position information, and transmitting a fusion result to the central control module;

the route guiding unit is used for screening out the lane information with the least passing vehicles according to the current image information of the parking lot;

and the parking space guiding unit is used for screening out empty parking spaces according to the current image information of the parking lot.

2. The intelligent parking positioning system based on UWB/IMU and visual information fusion of claim 1, wherein the environment error neural network learning model is an environment error perception deep learning model established by a convolutional neural network, and error factors of positioning deviation caused by environment factors are extracted and combined layer by layer to generate high-level features in an abstract manner, so as to help a UWB/IMU module to correct positioning accuracy; the method for extracting the error factor of the positioning deviation caused by the environmental factors comprises the following steps:

nth fixed UWB base station coordinate U _n ＝(x _n ,y _n ,z _n ) Known coordinates; the position of the vehicle to be positioned at the moment t is recorded as N _t ＝(x _t ,y _t ,z _t ) (ii) a the distance from the UWB base station to the target vehicle at the time t is as follows:

wherein

For this purpose, an error factor;

substituting error factors at different moments

Learning by an environment error neural network model; wherein

Is a high layerA characteristic amount;

in order to be the weight and the coefficient,

wherein v is the target vehicle travel speed; t is _i+1 、T _i Corresponding to the time recorded by a certain moment and a later time frame in the process of the traveling of the target vehicle, (T) _i+1 -T _i ) Is the time difference of travel; theta _i For this purpose, the vehicle wheel angle is calculated.

3. The intelligent parking positioning system based on UWB/IMU and visual information fusion of claim 1, wherein the vehicle information quantization unit is configured to quantize image information acquired by the camera unit into vehicle information, determine the number of vehicles in a current scene acquired by the camera unit, and calibrate a target vehicle, specifically:

the vehicle information quantization unit acquires vehicle image information, quantizes the image information into vehicle types, colors and license plate numbers of vehicles corresponding to the pixel points, sequentially generates unique character string codes for storage, and generates a vehicle number ID for each vehicle; the vehicle information quantization unit acquires a current time frame vehicle body image of a target vehicle, performs local image processing on the vehicle body image to obtain discrete pixel points corresponding to a vehicle body, a color and a license plate, converts the discrete pixel points into discrete quantity values, generates a unique vehicle digital ID within corresponding time, and performs identity calibration on the target vehicle.

4. The intelligent parking positioning system based on UWB/IMU and visual information fusion of claim 1, wherein the camera mechanism threshold unit is configured to determine a deflection angle threshold of the camera unit according to a number of vehicles in a current scene and control the camera unit to execute, and specifically:

the azimuth angle of the ith high-precision camera corresponding to the parking lot space coordinate (x, y, z) is (alpha) _i ,β _i ,γ _i ) The (x, y, z) pairsThe space coordinate position of the ith high-precision camera in the parking lot is acquired, and the number of target vehicles in the visual angle range is N according to the azimuth angle _k And constructing a state matrix equation:

wherein the camera has an analytic power of R _χ And xi is a threshold value set for the camera shooting mechanism, and when xi is less than or equal to N _k And when the camera mechanism threshold value unit sends a deflection instruction to the camera, so that the angular deflection of the camera is realized.

5. The intelligent parking positioning system based on the fusion of the UWB/IMU and the visual information as claimed in claim 1, wherein the UWB/IMU module is configured to obtain a distance between the target vehicle and the UWB base station and motion information of the vehicle, so as to obtain virtual coordinate information and an inertial heading direction of the target vehicle in the whole area of the parking lot, and specifically:

the UWB/IMU module acquires the distance between the target vehicle and the base station: space coordinates of each UWB base station in a parking lot are obtained, time transmitted by pulse signals between each UWB base station and a target vehicle is obtained, and the distance between each base station and the target vehicle is calculated, so that virtual coordinate information of the target vehicle in the parking lot can be obtained through calculation:

where m and n are used to identify different base stations, l _m,n Represents the distance between m and n UWB base stations, t being the pulse transmission time; c is the speed of light; (x, y, z) are virtual coordinates of the target vehicle within the parking lot;

the UWB/IMU module acquires the motion information of the vehicle: and acquiring accelerometer data E (epsilon) and gyroscope data E (sigma) through an IMU inertial module so as to obtain the inertial traveling direction of the target vehicle.

6. The intelligent parking positioning system based on UWB/IMU and visual information fusion of claim 1, wherein the data fusion module is configured to fuse the virtual coordinate information of the UWB/IMU module and the vehicle position information of the target vehicle of the camera unit, and transmit a fusion result to the central control module, specifically:

establishing a fusion target positioning optimization function:

G _i ＝f(x _i-1 ,u _i ,w _i )

H _i,j ＝h(y _j ,x _i ,v _i,j )

wherein E (epsilon) and E (sigma) are accelerometer data and gyroscope data measured by the MU inertia module,

coordinate data after UWB positioning processing, T is observation target vehicle time frame, w _i As vehicle response rate, G _i Equation of motion obtained for the camera unit tracking the target vehicle, H _i,j Trajectory prediction equation, u, determined for vehicle trajectory analysis module _i 、v _i,j To observe noise, x _i Is the target vehicle position, y _j The coordinates of the parking spaces are obtained;

and when the target positioning optimization function solves the minimum point, obtaining the real-time accurate position information of the finally optimized vehicle.

7. An intelligent parking positioning method based on UWB/IMU and visual information fusion is characterized in that, based on the intelligent parking positioning system based on UWB/IMU and visual information fusion as claimed in any one of claims 1-6, comprising the following steps:

the method comprises the following steps of S1, obtaining current image information of a parking lot through a camera unit, tracking a target vehicle and a scene where the vehicle is located in real time to obtain a motion equation, and determining vehicle position information of the target vehicle, wherein the current image information of the parking lot comprises vehicle image information, obstacle image information and image information of surrounding parking places and lane lines;

s2, quantizing the image information acquired by the camera unit into vehicle information through a vehicle information quantization unit, determining the number of vehicles in the current scene acquired by the camera unit, and calibrating the target vehicle;

s3, according to the number of vehicles in the current scene, determining a deflection angle threshold of a camera unit through a camera mechanism threshold unit and controlling the camera unit to execute the deflection angle threshold;

s4, the UWB/IMU module acquires the distance between the target vehicle and the UWB base station and the motion information of the vehicle, so that the virtual coordinate information and the inertial advancing direction of the target vehicle in the whole area of the parking lot are acquired;

s5, the vehicle track analysis module acquires track information of the target vehicle according to the vehicle image information at continuous moments, and transmits the track information to the central control module for learning of the environment error neural network learning model;

s6, establishing an environmental error perception deep learning model by a convolutional neural network in the environmental error neural network learning model, extracting an error factor of positioning deviation caused by environmental factors, and helping a UWB/IMU module and a camera unit to correct positioning accuracy;

s7, the signal transmission and processing module transmits the data of the UWB/IMU module and the camera unit to the central control module;

s8, the calculation and positioning display module performs coordinate calculation and visual position visual tracking display according to the data of the UWB/IMU module and the camera unit;

s9, the data fusion module fuses virtual coordinate information of the UWB/IMU module and vehicle position information of a target vehicle of the camera unit to obtain real-time accurate position information of the vehicle, and transmits a fusion result to the central control module;

s10, the path guiding unit screens out lane information with the least passing vehicles according to the current image information of the parking lot, the parking space guiding unit screens out empty parking spaces according to the current image information of the parking lot, and the central control module guides the target vehicle to park according to the real-time accurate position information of the vehicle, the lane information and the coordinates of the empty parking spaces.

8. An intelligent parking positioning method based on UWB/IMU and visual information fusion as claimed in claim 7, wherein in step S6, the manner of extracting the error factor of the positioning deviation caused by the environmental factors includes:

nth fixed UWB base station coordinate U _n ＝(x _n ,y _n ,z _n ) Known coordinates; the position of the vehicle to be positioned at the moment t is recorded as N _t ＝(x _t ,y _t ,z _t ) (ii) a the distance from the UWB base station to the target vehicle at the time t is as follows:

wherein

For this purpose, an error factor;

substituting error factors at different moments

Learning by an environment error neural network model; wherein

Is a high level feature quantity;

in order to be the weight and the coefficient,

wherein v is the target vehicle travel speed; t is _i+1 、T _i Time (T) recorded corresponding to a certain moment and a later time frame in the process of moving the target vehicle _i+1 -T _i ) Is the time difference of travel; theta.theta. _i For this purpose, the vehicle is steered.

9. The intelligent parking positioning method based on UWB/IMU and visual information fusion of claim 7, wherein the step S3 specifically comprises:

the azimuth angle of the ith high-precision camera corresponding to the parking lot space coordinate (x, y, z) is (alpha) _i ,β _i ,γ _i ) And (x, y, z) corresponds to the spatial coordinate position of the ith high-precision camera in the parking lot, and the number N of target vehicles in the view angle range is acquired according to the azimuth angle _k And constructing a state matrix equation:

wherein the camera has an analytic power of R _χ And xi is a threshold value set for the camera shooting mechanism, and when xi is less than or equal to N _k And when the camera mechanism threshold value unit sends a deflection instruction to the camera, so that the angular deflection of the camera is realized.

10. The intelligent parking positioning method based on UWB/IMU and visual information fusion of claim 7, wherein the step S9 is specifically:

establishing a fusion target positioning optimization function:

G _i ＝f(x _i-1 ,u _i ,w _i )

H _i,j ＝h(y _j ,x _i ,v _i,j )

wherein E (epsilon) and E (sigma) are accelerometer data and gyroscope data measured by the MU inertia module,

handling squats for UWB positioningTarget data, T is the observation target vehicle time frame, w _i As vehicle response rate, G _i Equation of motion obtained for the camera unit to track the target vehicle, H _i,j Trajectory prediction equation, u, determined for a vehicle trajectory analysis module _i 、v _i,j To observe noise, x _i Is the target vehicle position, y _j The coordinates of the parking spaces;

and when the target positioning optimization function solves the minimum point, the real-time accurate position information of the finally optimized vehicle is obtained.

Images