Auxiliary positioning system based on computer vision

Abstract

The invention discloses an auxiliary positioning system based on computer vision. According to the system, through cooperative work of the information acquisition and data input module, the preliminary measurement calibration module, the secondary precision calibration module, the comprehensive evaluation and decision module and the dynamic adjustment and feedback module, high-precision position estimation is realized. According to the system, firstly, data such as position, speed and acceleration are collected through a sensor, and preliminary calibration is performed in combination with environmental parameters; afterwards, errors are corrected through a secondary precision calibration module, and more accurate position estimation is generated. The comprehensive evaluation and decision module generates a final evaluation value according to a calibration result, and optimizes a control signal through the dynamic adjustment and feedback module, thereby guaranteeing the stability and high efficiency of the system in a complex environment. The system can be widely applied to the fields of automatic driving, robot navigation, intelligent monitoring and the like, and has relatively high environmental adaptability and positioning precision.

Description

Technical Field

[0001] The present invention relates to the technical field of computer vision, and particularly to an auxiliary positioning system based on computer vision. Background Art

[0002] With the continuous development of artificial intelligence and machine vision technologies, the application of computer vision in multiple fields has gradually increased. Especially in positioning systems, computer vision technology provides new possibilities for high-precision positioning. Traditional positioning systems mainly rely on GPS or inertial navigation systems, but these systems are easily interfered in complex environments, resulting in reduced positioning accuracy. In addition, existing vision-aided positioning systems usually have problems such as poor adaptability to environmental changes and difficulty in error correction.

[0003] Therefore, how to improve positioning accuracy and enable the system to adapt to different environmental changes has become an important topic in the current technological development. Summary of the Invention

[0004] The purpose of the present invention is to provide an auxiliary positioning system based on computer vision, which has strong environmental adaptability and high-precision positioning ability, and solves the problems of low accuracy and poor adaptability to environmental changes of existing positioning systems in complex environments.

[0005] To achieve the above purpose, the present invention provides the following technical solution: An auxiliary positioning system based on computer vision, the system includes the following modules: an information collection and data input module, a preliminary measurement and calibration module, a secondary accuracy calibration module, a comprehensive evaluation and decision-making module, and a dynamic adjustment and feedback module;

[0006] The information collection and data input module is responsible for collecting data from sensors and the environment to generate a preliminary input data set;

[0007] The preliminary measurement and calibration module is configured to perform preliminary calibration based on sensor data and environmental data to generate a preliminary position estimate;

[0008] The secondary accuracy calibration module is configured to further correct errors based on the preliminary calibration and in combination with external environmental conditions to obtain a more accurate calibration result;

[0009] The comprehensive evaluation and decision-making module is configured to generate a final evaluation value based on the preliminary and accurate calibration results and provide a basis for the generation of subsequent control signals;

[0010] The dynamic adjustment and feedback module is configured to generate a control signal based on the comprehensive evaluation result and the actual error.

[0011] Preferably, the information collection and data input module includes the following sub-modules:

[0012] Sensor data acquisition sub-module: Collect the position, speed, and acceleration data of the robot, convert them into a sensor data set D = {D 1 , D 2 ,... D n}, and provide it to the preliminary measurement and calibration module for use;

[0013] Environmental data acquisition sub-module: Collect environmental-related parameters including light intensity, temperature, and humidity, form environmental data H = {h 1 , h 2 ,... h m}, and transmit it to the subsequent preliminary measurement and calibration module.

[0014] Preferably, the preliminary measurement and calibration module includes the following sub-modules:

[0015] Preliminary measurement processing sub-module: Perform preliminary processing on the sensor data to obtain the average measurement value D avg , where D i is the measurement data of the i-th sensor; n is the number of sensor data;

[0016] Preliminary calibration calculation sub-module: According to the average value D avg of the sensor data output by the preliminary measurement processing sub-module and the environmental parameters H, input them into the position estimation algorithm to generate a preliminary position estimate P est ;

[0017] The expression of the position estimation algorithm is: P est = D avg ·f(H), where f(H) is an environmental parameter correction function that introduces environmental impacts into the calibration process, and its expression is: , where β i is the correction coefficient of the environmental parameter h i ; h i is the i-th environmental parameter.

[0018] Preferably, the secondary precision calibration module includes:

[0019] Error correction calculation sub-module: Further correct the preliminary calibration result based on the environmental data to generate an accurate calibration result P cal , where where h i is the i-th environmental factor; λ is the correction coefficient, indicating the influence degree of each environmental factor on the preliminary position estimate value;

[0020] When h i is relatively large, λ will be dynamically increased, thereby improving the sensitivity of calibration;

[0021] Conversely, when h i is small or unchanged, the correction effect of λ will be reduced.

[0022] Preferably, the comprehensive evaluation and decision-making module includes:

[0023] Weighted evaluation calculation sub-module: Weigh the preliminary calibration result and the precise calibration result to generate a comprehensive evaluation value I, I = α·P est +(1-α)·P cal ; where α is a weight factor dynamically adjusted by system feedback and is automatically adjusted according to the current operating state of the system;

[0024] When the system is in the precise operation state, α is small, making the precise calibration result P cal more dominant;

[0025] When the system is uncertain or varies greatly, the value of α will increase, increasing the influence of the preliminary calibration value P est on the evaluation value;

[0026] Nonlinear correction calculation sub-module: Use a nonlinear function to correct the weighted result to generate the final evaluation value I final I final = log(I + 1), perform logarithmic correction on the comprehensive evaluation value I to optimize the stability and accuracy of the result in a complex environment.

[0027] Preferably, the dynamic adjustment and feedback module includes:

[0028] Control signal generation sub-module: Generate a control signal S in the control signal generation algorithm according to the final evaluation value I final , and the control signal S control includes speed and steering angle control parameters; control S

[0029] S control = f(I final,error ), f(I final,error ) = γ·I final ·(1 + δ·error);

[0030] where γ is the initial gain factor of the control signal, δ is the error dynamic adjustment factor, and through the control signal generation algorithm, the system calculates the control signal according to the evaluation value and the current path error;

[0031] Control signal adjustment sub-module: Dynamically adjust the control signal according to the error in the actual driving process using the control signal adjustment algorithm;

[0032] where the control signal adjustment algorithm where ζ is the dynamic adjustment factor, which increases with the increase of the error and is set to strengthen the system's response to larger errors.

[0033] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0034] 1. By integrating the fusion calibration technology of sensor data and environmental data, the present invention achieves high positioning accuracy, especially significantly improving the accuracy in complex environments;

[0035] 2. The present invention proposes a dynamic calibration method based on environmental factor correction, which can real-time correct the initial position estimation error, enhancing the stability and adaptability of the system in dynamic environments;

[0036] 3. The present invention introduces a comprehensive evaluation and decision-making module, which generates a final evaluation value according to the initial and accurate calibration results, providing a reliable basis for subsequent dynamic adjustment and further improving the accuracy of the system;

[0037] 4. The dynamic adjustment and feedback module of the present invention can adjust the control signal according to the real-time error, thereby ensuring precise control during the positioning process and adapting to variable environmental factors. Description of the Drawings

[0038] Figure 1 is the module flow chart of the computer vision-based auxiliary positioning system of the present invention;

[0039] Figure 2 is the schematic diagram of the control signal generation and dynamic correction process of the present invention;

[0040] Figure 3 is the schematic diagram of the initial measurement calibration process of the present invention;

[0041] Figure 4 is the schematic diagram of the secondary accuracy calibration and correction process of the present invention. Detailed Embodiments

[0042] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0043] Please refer to Figures 1 to 4, The present invention provides a technical solution: an auxiliary positioning system based on computer vision. The auxiliary positioning system based on computer vision provided by the present invention aims to achieve high-precision and high-reliability positioning by integrating sensor data and environmental data and combining computer vision technology. The main task of the system is to gradually correct and improve the accuracy of position estimation according to sensor data and environmental parameters, ensuring that robots, autonomous vehicles or other intelligent devices can accurately and efficiently complete positioning tasks in complex environments.

[0044] The basic framework of the system includes an information collection and data input module, a preliminary measurement calibration module, a secondary accuracy calibration module, a comprehensive evaluation and decision-making module, and a dynamic adjustment and feedback module. Each module plays a key role in the system and cooperates with each other to continuously optimize the positioning accuracy.

[0045] 1. Information collection and data input module

[0046] The information collection and data input module is responsible for collecting the motion data (position, speed, acceleration) of the robot and environmental parameters (light intensity, temperature, humidity) from the environment. This module mainly includes two sub-modules: a sensor data collection sub-module and an environmental data collection sub-module.

[0047] The sensor data collection sub-module is responsible for collecting the position, speed, and acceleration motion data of the robot and organizing them into a data set D = {D 1 , D 2 ,... D n}, where D 1 , D 2 ,... D n represents multiple data points collected by the sensor, and n is the total number of sensor data. These data provide the basis for subsequent calibration and positioning calculations;

[0048] The environmental data collection sub-module is responsible for collecting environmental-related parameters such as light intensity, temperature, and humidity and organizing them into a data set H = {h 1 , h 2 ,... h m}, where h 1 , h 2 ,... h m represents various relevant parameters in the environment,

[0049] and m is the total number of environmental data. These environmental parameters will serve as important inputs for the calibration algorithm to help correct the errors of sensor data.

[0050] 2. Preliminary measurement calibration module

[0051] The main task of the preliminary measurement calibration module is to perform preliminary calibration using sensor data and environmental data to generate a preliminary position estimate. This module includes a preliminary measurement processing sub-module and a preliminary calibration calculation sub-module.

[0052] Preliminary measurement processing sub-module: First, perform preliminary processing on the collected sensor data D = {D 1 , D 2 ,... D n}, and calculate the average value D avg of the sensor data as the preliminary measurement value. This average value is calculated by the following formula: where D i is the measurement data of the i-th sensor; n is the number of sensor data, and D avg is the average value of all data points.

[0053] Preliminary calibration calculation sub-module: After obtaining the preliminary measurement value, the system combines the environmental data H = {h 1 , h 2 ,... h m} to perform preliminary calibration and generate a preliminary position estimate P est . The preliminary calibration is performed by inputting the average value D avg of the sensor data and the environmental data H into the calibration algorithm. This calibration algorithm can be expressed as: P est = D avg · f(H), where f(H) is the environmental parameter correction function that introduces environmental effects into the calibration process, and its expression is: where β i is the coefficient for correcting the sensor data by the environmental parameter h i ; h i is the i-th environmental parameter; m is the number of environmental parameters.

[0054] The correction coefficient β i of the environmental parameter h i can be determined through experiments or historical data, reflecting the degree of influence of environmental factors on sensor data.

[0055] 3. Secondary precision calibration module

[0056] The secondary precision calibration module is used to further correct the preliminary calibration result and provide a more accurate position estimate. The core of this module is the error correction calculation sub-module.

[0057] Error correction calculation sub-module: This module corrects the preliminary calibration result based on external environmental data to generate an accurate calibration result P cal , and the specific correction process is as follows: where h iis the i-th environmental factor; λ is a correction coefficient representing the influence degree of each environmental factor on the preliminary position estimation value;

[0058] When h i is relatively large, λ will be dynamically increased, thereby enhancing the calibration sensitivity;

[0059] Conversely, when h i is relatively small or unchanged, the correction effect of λ will be reduced.

[0060] The correction coefficient λ can be obtained through data analysis or system training and is used to quantify the influence of environmental factors on the preliminary position estimation.

[0061] 4. Comprehensive Evaluation and Decision-making Module

[0062] The main task of the comprehensive evaluation and decision-making module is to generate the final evaluation value I final . This module includes a weighted evaluation calculation sub-module and a non-linear correction calculation sub-module.

[0063] Weighted Evaluation Calculation Sub-module: According to the preliminary calibration result P est and the precise calibration result P cal , combined with the dynamically adjusted weight factor α, generate the comprehensive evaluation value I.

[0064] The weighted formula is as follows: I = α·P est +(1 - α)·P cal , where α is the weight factor dynamically adjusted by system feedback and is automatically adjusted according to the current operating state of the system. When the system is in an accurate operation state, α is relatively small, making the precise calibration result more dominant; while when the system is in an uncertain or highly variable state, α is relatively large, enhancing the influence of the preliminary calibration result.

[0065] Non-linear Correction Calculation Sub-module: Generate the final evaluation value I final by performing logarithmic correction on the weighted result I final to optimize the stability and accuracy of the system in a complex environment. The specific correction formula is: I

[0066] 5. Dynamic Adjustment and Feedback Module

[0067] The dynamic adjustment and feedback module is responsible for generating a control signal based on the comprehensive evaluation result and the path error, and dynamically adjusting the control signal to optimize the positioning performance of the system. This module includes a control signal generation sub-module and a control signal adjustment sub-module.

[0068] Control Signal Generation Sub-module: According to the final evaluation value Ifinal and the control algorithm generates a control signal S control , and the control signal includes parameters such as speed and steering angle. The formula for generating the control signal is: S control = f(I final,error ), f(I final,error ) = γ·I final ·(1 + δ·error), where γ is the initial gain factor of the control signal, and δ is the error dynamic adjustment factor, which reflects the influence of the path error on the control signal. This formula shows how the system adjusts the control signal according to the evaluation value and the path error to achieve precise navigation control. Through the control signal generation algorithm, the system calculates the control signal based on the evaluation value and the current path error;

[0069] Control signal adjustment sub-module: According to the error in the actual driving process, the control signal is dynamically adjusted using the control signal adjustment algorithm. The control signal adjustment formula is:

[0070] where ζ is the dynamic adjustment factor, which represents the influence of the path error on the control signal adjustment, increases with the increase of the error, and is set to enhance the system's response to larger errors.

[0071] Note: In S control = f(I final,error ) and , they are applied in sequence. The first formula calculates the preliminary signal, and the second formula performs feedback adjustment on it. In terms of the formula, they seem to be independent, but in fact, they are consecutive steps and do not directly conflict. Essentially, the second formula further corrects the S control generated by the first formula.

[0072] For example:

[0073] 1. The first formula: Assume S control = f(I final,error ), and a preliminary control signal is calculated.

[0074] 2. The second formula: Then this preliminary signal will be further adjusted by the second formula:

[0075] That is, the second formula modifies the output of the first formula, rather than performing two different operations on S control in two different directions.

[0076] Example 1: Application of a computer vision-based auxiliary positioning system in autonomous driving

[0077] With the rapid development of autonomous driving technology, accurate positioning and navigation have become the basis for the safe driving of driverless cars. Current autonomous driving systems mainly rely on GPS, inertial navigation systems (INS) and map data for positioning. However, in complex urban environments, underground parking lots or areas with weak GPS signals, the positioning accuracy of existing technologies is often affected, resulting in increased positioning errors, which in turn affects the safety and reliability of the system. Therefore, how to improve positioning accuracy in complex environments has become a key technical problem in autonomous driving systems.

[0078] The present invention aims to provide an auxiliary positioning system based on computer vision, which combines sensor data with environmental data, optimizes positioning accuracy through a multi-level calibration algorithm and a dynamic feedback adjustment mechanism, and can rely on environmental parameters for accurate calibration, especially in areas where GPS signals are weak or cannot be covered, thereby achieving high-precision autonomous driving positioning.

[0079] Multimodal data fusion: This invention combines sensor data (position, velocity, acceleration) and environmental data (light intensity, temperature, humidity) to optimize positioning results using a multi-level calibration algorithm. Unlike traditional systems that rely only on GPS or INS, this invention can achieve more accurate positioning by performing error correction through environmental data in areas where GPS signals are unavailable.

[0080] Dynamic adjustment and feedback mechanism: The present invention designs a dynamic adjustment and feedback module to dynamically adjust the control signal according to the comprehensive evaluation results and path error, ensuring that the system always maintains high accuracy and stability during actual operation.

[0081] Nonlinear correction method: In the comprehensive evaluation and decision-making module, a logarithmic correction algorithm is used to optimize high error conditions in complex environments, thereby improving the adaptability and stability of the system.

[0082] Autonomous driving in complex urban environments: In densely populated urban areas (such as high-rise buildings, tunnels, etc.), traditional positioning systems are prone to errors due to GPS signal reflection and occlusion. The present invention can effectively improve positioning accuracy and ensure the safe driving of autonomous vehicles in urban environments by supplementing and correcting environmental parameters.

[0083] Underground parking lots or areas without GPS signals: By combining sensor data with environmental data, the present invention can still maintain high-precision positioning when the GPS signal is weak or completely ineffective, ensuring that the vehicle can accurately navigate and complete parking operations.

[0084] Automatic parking system: During the automatic parking process, the vehicle needs to be parked accurately in a limited space. Since GPS signals are usually not covered in this environment, the present invention can improve the accuracy of automatic parking through multi-level calibration of sensors and environmental data.

[0085] Under different application scenarios, the present invention demonstrates obvious advantages. Especially in an environment where GPS signals are limited, it can effectively improve the positioning accuracy and ensure the safety and reliability of the autonomous driving system. The following is a comparison between the present invention and the prior art in several key indicators.

[0086] Comparative experiment

[0087]

[0088]

[0089] Explanation of experimental data:

[0090] In the area of high-rise buildings in the city, due to the reflection and occlusion of GPS signals by buildings, the positioning error of traditional GPS is large, up to 3.2 meters at most. However, by combining environmental data, light intensity, temperature, and humidity for correction, the positioning error of the present invention is reduced to 1.2 meters, and the accuracy is improved by 62.5%.

[0091] In the underground parking lot, traditional GPS cannot provide effective positioning information. However, the present invention realizes high-precision positioning of 0.5 meters through multi-level calibration of sensor data (acceleration, speed) and environmental data, far exceeding the traditional technology.

[0092] In an environment with good GPS signals such as on highways, the positioning accuracy of traditional technology and the present invention is not much different. However, in terms of accuracy optimization, the multi-level calibration algorithm of the present invention still provides an additional accuracy improvement with a promotion rate of 53.3%.

[0093] In the automatic parking scenario, due to weak signals, the positioning error of traditional GPS is relatively large, usually about 2.8 meters. In contrast, through the correction of sensor and environmental data, the accuracy of the present invention is greatly improved to 0.3 meters, and the positioning is more stable and accurate, almost eliminating the error.

[0094] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An auxiliary positioning system based on computer vision, characterized in that: The system includes the following modules: information collection and data input module, preliminary measurement calibration module, secondary precision calibration module, comprehensive evaluation and decision module and dynamic adjustment and feedback module; The information collection and data input module is responsible for collecting data from sensors and the environment and generating a preliminary input data set; The preliminary measurement calibration module is configured to perform preliminary calibration based on the sensor data and the environmental data to generate a preliminary position estimate; The secondary precision calibration module is configured to further correct errors based on the preliminary calibration and in combination with external environmental conditions to obtain a more accurate calibration result; The comprehensive evaluation and decision module is configured to generate a final evaluation value based on the preliminary and precise calibration results and provide a basis for subsequent control signal generation; The dynamic adjustment and feedback module is configured to generate a control signal according to the comprehensive evaluation result and the actual error.

2. The computer vision-based auxiliary positioning system according to claim 1, characterized in that: The information collection and data input module includes the following submodules: Sensor data acquisition submodule: collects the robot's position, velocity, and acceleration data and converts them into a sensor data set D = {D1, D2, ... D n }, and provide it to the preliminary measurement and calibration module; Environmental data collection submodule: collects parameters related to the environment, including light intensity, temperature, and humidity, to form environmental data H = {h1, h2, ... h m } and transfer it to the subsequent preliminary measurement calibration module.

3. The computer vision-based auxiliary positioning system according to claim 1, characterized in that: The preliminary measurement and calibration module includes the following submodules: Preliminary measurement processing submodule: Perform preliminary processing on sensor data to obtain the average measurement value D avg , Where D i is the i-th sensor measurement data; n is the number of sensor data; Preliminary calibration calculation submodule: Based on the average value D of the sensor data output by the preliminary measurement processing submodule avg and environmental parameters H, are input into the position estimation algorithm to generate a preliminary position estimate P est ; The expression of the position estimation algorithm is: P est =D avg f(H), where f(H) is the environmental parameter correction function, which introduces environmental influence into the calibration process, and its expression is: , where β i is the environmental parameter h i Correction factor; h i is the i-th environmental parameter.

4. The computer vision-based auxiliary positioning system according to claim 1, characterized in that: The secondary precision calibration module comprises: Error correction calculation submodule: further corrects the preliminary calibration results based on environmental data to generate accurate calibration results P cal ,in where h i is the i-th environmental factor; λ is the correction coefficient, which indicates the influence of each environmental factor on the preliminary position estimate; When h i When it is large, λ will be dynamically increased, thereby improving the sensitivity of the calibration; On the contrary, when h i When λ is small or constant, the correction effect of λ will be reduced.

5. The computer vision-based auxiliary positioning system according to claim 1, characterized in that: The comprehensive evaluation and decision-making module includes: Weighted evaluation calculation submodule: weights the preliminary calibration results and the precise calibration results to generate a comprehensive evaluation value I, I = α·P est +(1-α)·P cal ; where α is a weight factor dynamically adjusted by system feedback, which is automatically adjusted according to the current operating status of the system; When the system is in precise operation, α is small, making the precise calibration result P cal More dominant; When the system is uncertain or changes greatly, the value of α will increase, increasing the initial calibration value P est Impact on valuation; Nonlinear correction calculation submodule: Use nonlinear functions to correct the weighted results and generate the final evaluation value I final , I final =log(I+1), logarithmically correct the comprehensive evaluation value I to optimize the stability and accuracy of the results in complex environments.

6. The computer vision-based auxiliary positioning system according to claim 1, characterized in that: The dynamic adjustment and feedback module includes: Control signal generation submodule: According to the final evaluation value I final The control signal S is generated in the control signal generation algorithm. control , control signal S control Contains speed and steering angle control parameters; S control =f(I final,error ),f(I final,error )=γ·I final ·(1+δ·error); Where γ is the initial gain factor of the control signal, δ is the error dynamic adjustment factor, and through the control signal generation algorithm, the system calculates the control signal based on the evaluation value and the current path error; Control signal adjustment submodule: dynamically adjusts the control signal using the control signal adjustment algorithm according to the error in the actual driving process; The control signal adjustment algorithm where ζ is a dynamic adjustment factor that increases as the error increases and is set to strengthen the system response to larger errors.

Images