Intelligent obstacle avoidance method and system based on machine vision

Abstract

The invention relates to the field related to application of visual obstacle avoidance, discloses an intelligent obstacle avoidance method and system based on machine vision, and establishes a technical scheme of cooperation of an electronic virtual blind sidewalk and an electronic tactile stick by adopting a machine vision scheme of synchronizing map positioning data through a cloud. According to the intelligent blind sidewalk system, the situations that the basic construction and maintenance of the blind sidewalk in daily life are insufficient and the occupation is serious can be effectively avoided, the problem that the blind person goes out difficultly is solved, the accurate going-out guiding and obstacle avoiding effects can be achieved through multi-party interaction linkage of real-time scene vision, cloud map data and the user blind stick equipment end, and high safety guarantee is provided for going out of the blind person.

Description

Technical Field

[0001] The present invention relates to application-related fields of visual obstacle avoidance, and specifically to an intelligent obstacle avoidance method and system based on machine vision. Background Art

[0002] Visually impaired people refer to those with intellectual disabilities whose vision is impaired due to various reasons such as illness and they are unable to obtain information through vision. In order to enable such people to travel smoothly, the setting of blind paths is one of the most basic road facilities. The blind paths use different surface graphics to convey action signals such as moving forward or turning to the blind, thereby achieving the purpose of travel guidance.

[0003] However, in actual scenarios, facilities such as blind paths are seriously occupied. For some cities, the construction of blind paths is very random, and even fails to meet regulatory standards, and cannot provide support for blind people's travel. The navigation implementation method of existing navigation software cannot be used directly by the blind, because the blind cannot confirm the navigation content through vision, and cannot avoid obstacles or choose the correct road. Summary of the invention

[0004] The purpose of the present invention is to provide an intelligent obstacle avoidance method and system based on machine vision to solve the problems raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions: An intelligent obstacle avoidance method based on machine vision, comprising: Acquire machine vision data and corresponding device posture data, perform spatial analysis on the machine vision data based on the device posture data, and establish a spatial distribution model, wherein the spatial distribution model is used to characterize the distribution of the environment within the visible range; Synchronize map environment data based on current positioning data, map map environment data with spatial distribution model, and synchronize the passing requirements of the current area according to the navigation path; Performing path passability simulation on the mapped spatial distribution model to obtain an obstacle avoidance path for current passing requirements, and establishing an electronic virtual tactile path based on the obstacle avoidance path. The electronic virtual tactile path is distributed in a three-dimensional structure within the spatial distribution model, and its cross-section is larger than the user passing area. The device posture data and motion inertia data are synchronized in real time to calculate the position posture information of the tracking device in the spatial distribution model, and the relative position of the position posture information is judged through the electronic virtual blind path to generate and output a vibration feedback signal.

[0006] As a further solution of the present invention: the step of performing relative position judgment on the position and posture information through the electronic virtual blind path, generating a vibration feedback signal and outputting the signal specifically comprises: Based on the path direction, the electronic virtual blind path is continuously intercepted by length units, and each unit is optimized into a guide straight line, which is used to map the blind path bricks and is provided with directionality and feedback features; The position and posture information is judged according to the direction of the current guide line. If the orientation of the device is consistent with the guide line, an alignment vibration feedback is generated and output; The position and orientation of the device is determined based on the position and attitude information. When the device is located on both sides of the guide line, regression vibration feedback is generated and outputted respectively. The regression vibration feedback also includes vibration direction information, which is used to guide the device position correction. The feedback characteristic vibration signal of the guide line at the current position of the device is output at a preset time interval, and the feedback characteristic is used to characterize the guiding action of the current guide line, including straight-line action and turning action.

[0007] As a further solution of the present invention: it also includes the steps of: Establishing electronic walls at both sides of the electronic virtual blind path space, and when the device touches the electronic wall, boundary vibration feedback is triggered and output; If there is an obstacle object outside the electronic wall, a buffer feedback area is established inside the electronic wall according to the preset safety notice distance. When the device enters the buffer feedback area, the buffer notification feedback is triggered and output. The buffer notification feedback is used to notify the existence of the obstacle object, and the obstacle includes a conventional obstacle and the vertical space where the parallel lanes are located.

[0008] As a further solution of the present invention: the step of synchronizing the map environment data based on the current positioning data and mapping the map environment data with the spatial distribution model specifically includes: Acquire current positioning data, and make a data request to a map service based on the positioning data to obtain street view image information within a redundant distance before and after positioning, wherein the street view image information is used to represent a panoramic image of a road at a position corresponding to the current positioning data; Based on the orientation data, the map environment data and the spatial distribution model are basically superimposed, so that the orientation angle of the spatial distribution model is consistent with the orientation angle of the street view image information; Image feature retrieval is performed on the map environment data and the spatial distribution model to obtain a plurality of common image features, and position correction is performed on the current positioning data based on the plurality of common image features.

[0009] As a further solution of the present invention: it also includes the steps of: Performing blind path feature detection on the machine vision data of the spatial distribution model, if a blind path object is detected, setting the location of the blind path as a basic guidance path, and performing path passability simulation based on the basic guidance path according to the needs; If no blind path object is detected, the fixed obstacles in the spatial distribution model are identified, and a basic guidance path is established on the side of the fixed obstacles, wherein the fixed obstacles include walls and trees.

[0010] The embodiment of the present invention aims to provide an intelligent obstacle avoidance system based on machine vision, comprising: The environment twin module is used to obtain machine vision data and corresponding device posture data, perform spatial analysis on the machine vision data based on the device posture data, and establish a spatial distribution model, which is used to characterize the environment distribution within the visible range; A map mapping module is used to synchronize map environment data based on current positioning data, map the map environment data with the spatial distribution model, and synchronize the passing requirements of the current area according to the navigation path; The avoidance simulation module is used to simulate the path passability of the mapped spatial distribution model, obtain the obstacle avoidance path of the current passing demand, and establish an electronic virtual tactile path based on the obstacle avoidance path. The electronic virtual tactile path is distributed in a three-dimensional structure in the spatial distribution model, and the cross-section is larger than the user passing area; The guidance feedback module is used to synchronize the device posture data and motion inertia data in real time to calculate the position posture information of the tracking device in the spatial distribution model, and to make relative position judgment on the position posture information through the electronic virtual blind path, generate a vibration feedback signal and output it.

[0011] As a further solution of the present invention: the guidance feedback module includes: A segmented sharpening unit is used to perform continuous length unit interception on the electronic virtual blind path based on the path direction, and optimize each unit into a guide straight line, wherein the guide straight line is used to map the blind path bricks and is provided with directionality and feedback features; A direction judgment unit is used to judge the position and posture information according to the direction of the current guide line. If the orientation of the representation device is consistent with the guide line, an alignment vibration feedback is generated and output; A deviation judgment unit is used to judge the position and orientation of the device based on the position and posture information. When the device is located on both sides of the guide line, a regression vibration feedback is generated and outputted respectively. The regression vibration feedback also includes vibration direction information, which is used to guide the device position correction; The action feedback unit is used to output a feedback characteristic vibration signal of the guide line at the current position of the device at a preset time interval, and the feedback characteristic is used to characterize the guiding action of the current guide line, including a straight action and a turning action.

[0012] As a further solution of the present invention: it also includes a safety restriction module, specifically including: A boundary setting unit, used to establish electronic walls at two sides of the electronic virtual blind path space, and when the device touches the electronic wall, a boundary vibration feedback is triggered and output; The buffer setting unit is used to establish a buffer feedback area inside the electronic wall according to a preset safety notice distance if there is an obstacle object outside the electronic wall. When the device enters the buffer feedback area, the buffer notification feedback is triggered and output. The buffer notification feedback is used to notify the existence of the obstacle object, and the obstacle includes a conventional obstacle and the vertical space where the parallel lanes are located.

[0013] As a further solution of the present invention: the map mapping module includes: A street view acquisition unit, used to acquire current positioning data, and make a data request to a map service based on the positioning data to acquire street view image information within a redundant distance before and after positioning, wherein the street view image information is used to represent a road panoramic image at a position corresponding to the current positioning data; A basic overlay unit, used for performing basic overlay on the map environment data and the spatial distribution model based on the orientation data, so that the orientation angle of the spatial distribution model is consistent with the orientation angle of the street view image information; The position correction unit is used to perform image feature retrieval on the map environment data and the spatial distribution model, obtain multiple common image features, and perform position correction on the current positioning data based on the multiple common image features.

[0014] As a further solution of the present invention: it also includes: A blind path guiding unit is used to detect blind path features on the machine vision data of the spatial distribution model. If a blind path object is detected, the location of the blind path is set as a basic guiding path, and path passability simulation is performed based on the basic guiding path according to the requirements. The obstacle guidance unit is used to identify the fixed obstacles in the spatial distribution model if no blind path object is detected, and establish a basic guidance path on the side of the fixed obstacles, wherein the fixed obstacles include walls and trees.

[0015] Compared with the prior art, the beneficial effects of the present invention are: a machine vision solution that synchronizes map positioning data through the cloud is adopted to establish a technical solution for the coordination of electronic virtual blind paths and electronic blind sticks, which can effectively avoid the lack of maintenance and serious occupancy of daily blind path infrastructure, and solve the problem of blind people's travel difficulties. The multi-party interactive linkage between real-time scene vision, cloud map data and the user's blind stick device can achieve accurate travel guidance and obstacle avoidance effects, providing a higher safety guarantee for blind people's travel. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 The figure is a flowchart of an intelligent obstacle avoidance method based on machine vision.

[0017] Figure 2 A flowchart of the steps for generating vibration feedback in an intelligent obstacle avoidance method based on machine vision.

[0018] Figure 3 This is a block diagram of the composition of an intelligent obstacle avoidance system based on machine vision. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0020] The specific implementation of the present invention is described in detail below in conjunction with specific embodiments.

[0021] like Figure 1 The present invention provides an intelligent obstacle avoidance method based on machine vision, which comprises the following steps: S10, acquiring machine vision data and corresponding device posture data, performing spatial analysis on the machine vision data based on the device posture data, and establishing a spatial distribution model, wherein the spatial distribution model is used to characterize the distribution of the environment within the visible range; S20, synchronizing map environment data based on current positioning data, mapping the map environment data with the spatial distribution model, and synchronizing the passing requirements of the current area according to the navigation path; S30, performing path passability simulation on the mapped spatial distribution model, obtaining an obstacle avoidance path for current passing requirements, and establishing an electronic virtual tactile path based on the obstacle avoidance path, wherein the electronic virtual tactile path is distributed in a three-dimensional structure in the spatial distribution model, and the cross-section is larger than the user passing area; S40, synchronizing the device posture data and motion inertia data in real time to calculate the position posture information of the tracking device in the spatial distribution model, and performing relative position judgment on the position posture information through the electronic virtual blind path, generating and outputting a vibration feedback signal.

[0022] In this embodiment, an intelligent obstacle avoidance method based on machine vision is provided. A machine vision solution is adopted to synchronize map positioning data through the cloud, and a technical solution for the coordination of an electronic virtual blind path and an electronic blind stick is established. This can effectively avoid the lack of maintenance and serious occupation of the basic infrastructure of the blind path in daily life, and solve the problem of difficulty in blind travel. The multi-party interaction between real-time scene vision, cloud map data and the user's blind stick device can achieve accurate travel guidance and obstacle avoidance effects, providing a higher safety guarantee for blind travel. In order to ensure that the blind can travel independently and safely, infrastructure such as blind paths that facilitate blind travel is usually built during municipal construction, but In real life, the occupation of facilities such as blind paths is serious. Even in some cities, the construction of blind paths is very random and fails to meet regulatory standards, and cannot provide support for blind people's travel. The navigation implementation method of existing navigation software cannot be directly used by the blind, because the blind cannot confirm the navigation content through vision, and cannot avoid obstacles and choose the correct road. The technical solution adopted in this embodiment can be summarized as follows: by setting a machine vision-related structure for the blind stick and linking data with the navigation software cloud, the establishment of an electronic virtual blind path is realized, that is, the environment is visually recognized by the software, and simulated obstacle avoidance is performed. , in order to establish a virtual non-existent electronic blind path, so as to exchange information with the user through the vibration feedback of the blind stick (simulating the information interaction method when the blind stick slides and knocks on the ground). In this way, the direct information exchange between the existing navigation software navigation method and the blind user is avoided, and a more efficient and streamlined information interaction method simulating the traditional blind stick is adopted. For the blind, it is easier to quickly obtain streamlined and effective guidance information; specifically, the machine vision data (i.e., image data) is acquired by the image acquisition device, and then the spatial distribution model within the visible range of the image acquisition device can be established, and then the positioning data and posture data (characterizing the spatial distribution model) are used. The spatial distribution model is mapped with the map (the southeast, northwest, and north directions of the model), and the map contains navigation data to the destination. Therefore, the current position in the spatial distribution model and the target position at a certain distance can be synchronized according to the navigation data (that is, a passing requirement is established), and then an obstacle avoidance simulation is performed between these two points to generate an electronic virtual blind path that can avoid obstacles (including aerial obstacles within a certain height range). When the user swings the blind stick, the position information of the blind stick in the electronic virtual blind path can be updated in real time based on the attitude sensor, inertial sensor, etc., so as to provide information feedback to the user by simulating different vibration modes, thereby achieving the purpose of guidance.

[0023] like Figure 2 As shown, as another preferred embodiment of the present invention, the step of performing relative position judgment on the position posture information through the electronic virtual blind path, generating a vibration feedback signal and outputting it specifically includes: S41, based on the path direction, the electronic virtual tactile path is continuously intercepted into length units, and each unit is optimized into a guide straight line, wherein the guide straight line is used to map the tactile path bricks and is provided with directionality and feedback features; S42, judging the position and posture information according to the direction of the current guide straight line, and if the orientation of the representation device is consistent with the guide straight line, generating and outputting alignment vibration feedback; S43, determining the position and orientation of the device based on the position and attitude information, and generating and outputting corresponding regression vibration feedback when the device is located on both sides of the guide line, wherein the regression vibration feedback also includes vibration direction information, which is used to guide the device to correct its position; S44, outputting a feedback characteristic vibration signal of the guide line at the current position of the device at a preset time interval, wherein the feedback characteristic is used to characterize the guiding action of the current guide line, including a straight action and a turning action.

[0024] In this embodiment, several vibration feedback signal output conditions in common scenarios are added. First, the road needs to be segmented and straightened, because a road that is too curved and complex is complicated and unfavorable for a blind person who uses a blind stick to grope forward. The limited information density obtained by the blind stick cannot complete the purpose of complex curved travel. Therefore, the route needs to be fragmented and straightened, just like a piece of bricks on a blind path. The first is the judgment of direction. By aligning the vibration feedback when the direction of the blind stick is consistent with the guiding straight line, the blind person can be quickly informed of the correctness of the direction of travel. The second is the judgment of the center of the route position. When the blind person leans to one side of the electronic virtual blind path, the risk of encountering or colliding with obstacles will increase. Therefore, the blind person is informed of which side to move to the left or right to return to the center of the blind path through directional vibration simulation (i.e., regression vibration feedback); the third is to tell the blind person through different vibrations whether to continue moving forward or prepare to turn at this time, so that the blind person can more quickly determine the trial method when using the blind stick to try.

[0025] As another preferred embodiment of the present invention, the steps are also included: Establishing electronic walls at both sides of the electronic virtual blind path space, and when the device touches the electronic wall, boundary vibration feedback is triggered and output; If there is an obstacle object outside the electronic wall, a buffer feedback area is established inside the electronic wall according to the preset safety notice distance. When the device enters the buffer feedback area, the buffer notification feedback is triggered and output. The buffer notification feedback is used to notify the existence of the obstacle object, and the obstacle includes a conventional obstacle and the vertical space where the parallel lanes are located.

[0026] In this embodiment, in the actual travel process, there will be various complex roadside obstacles. During the travel, the large-scale swinging and knocking of the blind stick may hit passers-by or placed fragile items, causing unnecessary trouble. Therefore, an electronic wall and a buffer feedback area are set here to provide action stop notifications and advance obstacle existence notifications, so that users can use the blind stick in a suitable space to avoid unnecessary trouble for themselves.

[0027] As another preferred embodiment of the present invention, the step of synchronizing the map environment data based on the current positioning data and mapping the map environment data with the spatial distribution model specifically includes: Acquire current positioning data, and make a data request to a map service based on the positioning data to obtain street view image information within a redundant distance before and after positioning, wherein the street view image information is used to represent a panoramic image of a road at a position corresponding to the current positioning data; Based on the orientation data, the map environment data and the spatial distribution model are basically superimposed, so that the orientation angle of the spatial distribution model is consistent with the orientation angle of the street view image information; Image feature retrieval is performed on the map environment data and the spatial distribution model to obtain a plurality of common image features, and position correction is performed on the current positioning data based on the plurality of common image features.

[0028] In this embodiment, the map navigation method in the prior art requires the user to confirm the information through vision to complete the navigation purpose, and in navigation, the positioning cannot be 100% accurate. For the blind, it is impossible to confirm through vision, and the inaccurate navigation positioning will have a certain impact on their obstacle avoidance process. There is a certain deviation between the position of the actual collected image data and the position in the map navigation, which may affect the accurate establishment of the subsequent electronic virtual blind path. Therefore, a positioning correction method that cooperates with the panoramic street view of the map software is supplemented here, that is, image feature fitting is performed through machine vision data and street view image information. Because the street view image information is collected by a fixed-point collection vehicle, its positioning is absolutely accurate (ignoring a small amount of error in the software modeling process). Therefore, by fitting the array machine vision data in the spatial distribution model, the specific position of the current collection point can be accurately calculated, and the positioning calibration process is completed to ensure the accuracy of the subsequent establishment of the electronic virtual blind path.

[0029] As another preferred embodiment of the present invention, the steps are also included: Performing blind path feature detection on the machine vision data of the spatial distribution model, if a blind path object is detected, setting the location of the blind path as a basic guidance path, and performing path passability simulation based on the basic guidance path according to the needs; If no blind path object is detected, the fixed obstacles in the spatial distribution model are identified, and a basic guidance path is established on the side of the fixed obstacles, wherein the fixed obstacles include walls and trees.

[0030] In this embodiment, an auxiliary process for establishing an electronic virtual blind path is added. For some roads, the width may be relatively large. Therefore, when establishing the electronic virtual blind path, certain setting standards need to be adopted to avoid the location of the blind path being too close to unsafe areas (for example, close to lanes, or areas with complex pedestrian flow, where various unassessable sudden intrusions may occur when users pass through these areas). Therefore, when establishing the electronic virtual blind path, if a blind path already exists, it is established based on the original physical blind path. The existing blind path is usually set up in a safe position close to the inside and away from vehicles. If there is no existing blind path, the best solution is to set it up close to fences, walls, and roadside trees, which can effectively avoid sudden intrusions on one side, reduce the probability of vehicles appearing, and improve travel safety.

[0031] like Figure 3 As shown, the present invention also provides an intelligent obstacle avoidance system based on machine vision, which comprises: The environment twin module 100 is used to obtain machine vision data and corresponding device posture data, perform spatial analysis on the machine vision data based on the device posture data, and establish a spatial distribution model, which is used to characterize the environment distribution within the visible range; A map mapping module 200 is used to synchronize map environment data based on current positioning data, map the map environment data with a spatial distribution model, and synchronize the passing requirements of the current area according to the navigation path; The avoidance simulation module 300 is used to simulate the path passability of the mapped spatial distribution model, obtain the obstacle avoidance path of the current passing demand, and establish an electronic virtual tactile path based on the obstacle avoidance path. The electronic virtual tactile path is distributed in a three-dimensional structure in the spatial distribution model, and the cross-section is larger than the user passing area; The guidance feedback module 400 is used to synchronize the device posture data and motion inertia data in real time to calculate the position posture information of the tracking device in the spatial distribution model, and to perform relative position judgment on the position posture information through the electronic virtual blind path, generate a vibration feedback signal and output it.

[0032] As another preferred embodiment of the present invention, the guidance feedback module includes: A segmented sharpening unit is used to perform continuous length unit interception on the electronic virtual blind path based on the path direction, and optimize each unit into a guide straight line, wherein the guide straight line is used to map the blind path bricks and is provided with directionality and feedback features; A direction judgment unit is used to judge the position and posture information according to the direction of the current guide line. If the orientation of the representation device is consistent with the guide line, an alignment vibration feedback is generated and output; A deviation judgment unit is used to judge the position and orientation of the device based on the position and posture information. When the device is located on both sides of the guide line, a regression vibration feedback is generated and outputted respectively. The regression vibration feedback also includes vibration direction information, which is used to guide the device position correction; The action feedback unit is used to output a feedback characteristic vibration signal of the guide line at the current position of the device at a preset time interval, and the feedback characteristic is used to characterize the guiding action of the current guide line, including a straight action and a turning action.

[0033] As another preferred embodiment of the present invention, it also includes a safety restriction module, specifically including: A boundary setting unit, used to establish electronic walls at two sides of the electronic virtual blind path space, and when the device touches the electronic wall, a boundary vibration feedback is triggered and output; The buffer setting unit is used to establish a buffer feedback area inside the electronic wall according to a preset safety notice distance if there is an obstacle object outside the electronic wall. When the device enters the buffer feedback area, the buffer notification feedback is triggered and output. The buffer notification feedback is used to notify the existence of the obstacle object, and the obstacle includes a conventional obstacle and the vertical space where the parallel lanes are located.

[0034] As another preferred embodiment of the present invention, the map mapping module includes: A street view acquisition unit, used to acquire current positioning data, and make a data request to a map service based on the positioning data to acquire street view image information within a redundant distance before and after positioning, wherein the street view image information is used to represent a road panoramic image at a position corresponding to the current positioning data; A basic overlay unit, used for performing basic overlay on the map environment data and the spatial distribution model based on the orientation data, so that the orientation angle of the spatial distribution model is consistent with the orientation angle of the street view image information; The position correction unit is used to perform image feature retrieval on the map environment data and the spatial distribution model, obtain multiple common image features, and perform position correction on the current positioning data based on the multiple common image features.

[0035] As another preferred embodiment of the present invention, it also includes: A blind path guiding unit is used to detect blind path features on the machine vision data of the spatial distribution model. If a blind path object is detected, the location of the blind path is set as a basic guiding path, and path passability simulation is performed based on the basic guiding path according to the requirements. The obstacle guidance unit is used to identify the fixed obstacles in the spatial distribution model if no blind path object is detected, and establish a basic guidance path on the side of the fixed obstacles, wherein the fixed obstacles include walls and trees.

[0036] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0037] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the disclosure in the specification and examples. This application is intended to cover any variations, uses or adaptations of the present disclosure, which follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present disclosure are indicated by the claims.

[0038] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. An intelligent obstacle avoidance method based on machine vision, characterized in that: Include: Acquire machine vision data and corresponding device posture data, perform spatial analysis on the machine vision data based on the device posture data, and establish a spatial distribution model, wherein the spatial distribution model is used to characterize the distribution of the environment within the visible range; Synchronize map environment data based on current positioning data, map map environment data with spatial distribution model, and synchronize the passing requirements of the current area according to the navigation path; Performing path passability simulation on the mapped spatial distribution model to obtain an obstacle avoidance path for current passing requirements, and establishing an electronic virtual tactile path based on the obstacle avoidance path. The electronic virtual tactile path is distributed in a three-dimensional structure within the spatial distribution model, and its cross-section is larger than the user passing area. The device posture data and motion inertia data are synchronized in real time to calculate the position posture information of the tracking device in the spatial distribution model, and the relative position of the position posture information is judged through the electronic virtual blind path to generate and output a vibration feedback signal.

2. The intelligent obstacle avoidance method based on machine vision according to claim 1, characterized in that: The step of performing relative position judgment on the position and posture information by using the electronic virtual blind path, generating a vibration feedback signal and outputting the signal specifically comprises: Based on the path direction, the electronic virtual blind path is continuously intercepted by length units, and each unit is optimized into a guide straight line, which is used to map the blind path bricks and is provided with directionality and feedback features; The position and posture information is judged according to the direction of the current guide line. If the orientation of the device is consistent with the guide line, an alignment vibration feedback is generated and output; The position and orientation of the device is determined based on the position and attitude information. When the device is located on both sides of the guide line, regression vibration feedback is generated and outputted respectively. The regression vibration feedback also includes vibration direction information, which is used to guide the device position correction. The feedback characteristic vibration signal of the guide line at the current position of the device is output at a preset time interval, and the feedback characteristic is used to characterize the guiding action of the current guide line, including straight-line action and turning action.

3. The intelligent obstacle avoidance method based on machine vision according to claim 1, characterized in that: Also includes the steps: Establishing electronic walls at both sides of the electronic virtual blind path space, and when the device touches the electronic wall, boundary vibration feedback is triggered and output; If there is an obstacle object outside the electronic wall, a buffer feedback area is established inside the electronic wall according to the preset safety notice distance. When the device enters the buffer feedback area, the buffer notification feedback is triggered and output. The buffer notification feedback is used to notify the existence of the obstacle object, and the obstacle includes a conventional obstacle and the vertical space where the parallel lanes are located.

4. The intelligent obstacle avoidance method based on machine vision according to claim 1, characterized in that: The step of synchronizing the map environment data based on the current positioning data and mapping the map environment data with the spatial distribution model specifically includes: Acquire current positioning data, and make a data request to a map service based on the positioning data to obtain street view image information within a redundant distance before and after positioning, wherein the street view image information is used to represent a panoramic image of a road at a position corresponding to the current positioning data; Based on the orientation data, the map environment data and the spatial distribution model are basically superimposed, so that the orientation angle of the spatial distribution model is consistent with the orientation angle of the street view image information; Image feature retrieval is performed on the map environment data and the spatial distribution model to obtain a plurality of common image features, and position correction is performed on the current positioning data based on the plurality of common image features.

5. The intelligent obstacle avoidance method based on machine vision according to claim 3, characterized in that: Also includes the steps: Performing blind path feature detection on the machine vision data of the spatial distribution model, if a blind path object is detected, setting the location of the blind path as a basic guidance path, and performing path passability simulation based on the basic guidance path according to the needs; If no blind path object is detected, the fixed obstacles in the spatial distribution model are identified, and a basic guidance path is established on the side of the fixed obstacles, wherein the fixed obstacles include walls and trees.

6. An intelligent obstacle avoidance system based on machine vision, characterized in that: Include: The environment twin module is used to obtain machine vision data and corresponding device posture data, perform spatial analysis on the machine vision data based on the device posture data, and establish a spatial distribution model, which is used to characterize the environment distribution within the visible range; A map mapping module is used to synchronize map environment data based on current positioning data, map the map environment data with the spatial distribution model, and synchronize the passing requirements of the current area according to the navigation path; The avoidance simulation module is used to simulate the path passability of the mapped spatial distribution model, obtain the obstacle avoidance path of the current passing demand, and establish an electronic virtual tactile path based on the obstacle avoidance path. The electronic virtual tactile path is distributed in a three-dimensional structure in the spatial distribution model, and the cross-section is larger than the user passing area; The guidance feedback module is used to synchronize the device posture data and motion inertia data in real time to calculate the position posture information of the tracking device in the spatial distribution model, and to make relative position judgment on the position posture information through the electronic virtual blind path, generate a vibration feedback signal and output it.

7. The intelligent obstacle avoidance system based on machine vision according to claim 6, characterized in that: The guidance feedback module comprises: A segmented sharpening unit is used to perform continuous length unit interception on the electronic virtual blind path based on the path direction, and optimize each unit into a guide straight line, wherein the guide straight line is used to map the blind path bricks and is provided with directionality and feedback features; A direction judgment unit is used to judge the position and posture information according to the direction of the current guide line. If the orientation of the representation device is consistent with the guide line, an alignment vibration feedback is generated and output; A deviation judgment unit is used to judge the position and orientation of the device based on the position and posture information. When the device is located on both sides of the guide line, a regression vibration feedback is generated and outputted respectively. The regression vibration feedback also includes vibration direction information, which is used to guide the device position correction; The action feedback unit is used to output a feedback characteristic vibration signal of the guide line at the current position of the device at a preset time interval, and the feedback characteristic is used to characterize the guiding action of the current guide line, including a straight action and a turning action.

8. The intelligent obstacle avoidance system based on machine vision according to claim 6, characterized in that: It also includes safety restriction modules, including: A boundary setting unit, used to establish electronic walls at two sides of the electronic virtual blind path space, and when the device touches the electronic wall, a boundary vibration feedback is triggered and output; The buffer setting unit is used to establish a buffer feedback area inside the electronic wall according to a preset safety notice distance if there is an obstacle object outside the electronic wall. When the device enters the buffer feedback area, the buffer notification feedback is triggered and output. The buffer notification feedback is used to notify the existence of the obstacle object, and the obstacle includes a conventional obstacle and the vertical space where the parallel lanes are located.

9. The intelligent obstacle avoidance system based on machine vision according to claim 6, characterized in that: The map mapping module includes: A street view acquisition unit, used to acquire current positioning data, and make a data request to a map service based on the positioning data to acquire street view image information within a redundant distance before and after positioning, wherein the street view image information is used to represent a road panoramic image at a position corresponding to the current positioning data; A basic overlay unit, used for performing basic overlay on the map environment data and the spatial distribution model based on the orientation data, so that the orientation angle of the spatial distribution model is consistent with the orientation angle of the street view image information; The position correction unit is used to perform image feature retrieval on the map environment data and the spatial distribution model, obtain multiple common image features, and perform position correction on the current positioning data based on the multiple common image features.

10. The intelligent obstacle avoidance system based on machine vision according to claim 8, characterized in that: Also includes: A blind path guiding unit is used to detect blind path features on the machine vision data of the spatial distribution model. If a blind path object is detected, the location of the blind path is set as a basic guiding path, and path passability simulation is performed based on the basic guiding path according to the requirements. The obstacle guidance unit is used to identify the fixed obstacles in the spatial distribution model if no blind path object is detected, and establish a basic guidance path on the side of the fixed obstacles, wherein the fixed obstacles include walls and trees.

Images