Blind sidewalk obstacle detection system

Abstract

The invention relates to the technical field of road traffic, and discloses a blind sidewalk obstacle detection system comprising a vehicle body which is internally provided with a detection module, an acquisition module, a GPS positioning module, an auxiliary module, a calculation module, a control module, an Internet of Things module and a power module; the detection module is used for detecting whether obstacles exist on the blind sidewalk; the acquisition module is used for acquiring images of blind sidewalks and obstacles; the GPS positioning module is used for acquiring a real-time position; the auxiliary module is used for detecting a vehicle body steering angle; the calculation module is used for obstacle detection, path planning and motion control; the control module is used for outputting a motion control strategy. According to the invention, the technical effects of real-time detection and accurate recognition of the obstacles on the blind sidewalk are achieved by collecting the ultrasonic sensor and image data, so that the accuracy of obstacle detection can be greatly improved through cooperative work of multiple sensors, and the problem of missing detection or misjudgment of the obstacles on the blind sidewalk caused by a single detection mode is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of road traffic, and particularly to a blind path obstacle detection system. Background Art

[0002] A blind path, also known as a blind path facility, is a dedicated road or guiding system provided for visually impaired people, usually laid on sidewalks or public places. Through the raised strip signs or special ground designs on the ground, blind or visually impaired people can determine the traveling direction through tactile perception, so as to walk safely or carry out other daily activities. The design of the blind path not only considers the convenience of walking, but also ensures the mobility and independence of visually impaired people in the urban environment.

[0003] The existence of blind path obstacles poses a great threat to the safety of visually impaired people. When there are obstacles on the blind path, visually impaired people may not be able to perceive them in time, resulting in accidents such as falling or collision. For the users of the blind path, keeping the blind path unobstructed is the basic premise to ensure their walking safety. Therefore, timely and accurately detecting obstacles on the blind path is of great significance for reducing accidents and improving the use efficiency of the blind path.

[0004] In the prior art, most blind path obstacle detection systems rely on a single sensor to detect obstacles. However, this method has obvious deficiencies in practical applications. For example, although traditional ultrasonic sensors can detect obstacles ahead, their detection range and accuracy are limited, and they are easily affected by the environment, resulting in missed detection or misjudgment of blind path obstacles. At the same time, a single vision sensor is also affected by factors such as lighting conditions and viewing angles, and cannot provide real-time and comprehensive obstacle information. Therefore, the prior art has certain limitations in blind path obstacle detection. Summary of the Invention

[0005] Aiming at the deficiencies of the prior art, the present invention provides a blind path obstacle detection system, which solves the problems of missed detection, misjudgment and affected accuracy by environmental interference in the prior art due to the dependence on a single sensor for blind path obstacle detection.

[0006] To achieve the above object, the present invention is realized through the following technical solutions: A blind path obstacle detection system, comprising:

[0007] A vehicle body, inside which a detection module, a collection module, a GPS positioning module, an auxiliary module, a calculation module, a control module, an Internet of Things module and a power module are installed;

[0008] The detection module includes an ultrasonic sensor for detecting whether there are obstacles on the blind path;

[0009] The acquisition module includes a camera, which is connected to the vehicle body through a pan-tilt head. The camera is used to collect images of the blind path and obstacles.

[0010] The GPS positioning module is used to obtain the real-time position.

[0011] The auxiliary module includes a gyroscope, which is used to detect the steering angle of the vehicle body.

[0012] The calculation module is connected to the ultrasonic sensor, camera, GPS positioning module and gyroscope. The calculation module performs obstacle detection, path planning and motion control based on the distance data provided by the ultrasonic sensor, the image data provided by the camera, the position data provided by the GPS positioning module, and the steering angle provided by the gyroscope.

[0013] The control module is used to output a motion control strategy according to the path planning and motion control.

[0014] The power module is used to drive the vehicle body to move according to the motion control strategy.

[0015] The Internet of Things module is used to transmit the real-time position and obstacle detection information to the user terminal.

[0016] Preferably, the calculation module includes an image processing unit. The image processing unit performs blind path obstacle recognition based on the YOLOV8 algorithm. The YOLOV8 algorithm is used to identify the obstacles in the image, classify them according to their categories, and adjust the motion control strategy of the control module according to the categories.

[0017] Preferably, the calculation module performs time series analysis on the image data based on the LSTM algorithm to identify and predict the motion trajectories of dynamic obstacles.

[0018] Preferably, the calculation module generates a path map of the blind path based on the graph theory algorithm and uses the Dijkstra algorithm to calculate the shortest path from the current position information to the target position. The control module adjusts the motion trajectory of the vehicle body 1 according to the shortest path.

[0019] Preferably, the calculation module performs path optimization based on the ant colony algorithm. The ant colony algorithm updates the pheromone of the path according to the influence degree of obstacles on each path. The control unit adjusts the motion trajectory of the vehicle body 1 according to the optimized path.

[0020] Preferably, when the calculation module detects a break in the blind path, it judges whether the blind path has broken by analyzing the changes between the image data provided by the camera and the historical image data, and sends the real-time position information to the user terminal through the Internet of Things module when a break occurs.

[0021] Preferably, when the ultrasonic sensor detects an obstacle, the calculation module controls the pan-tilt to adjust the angle of the camera for obstacle recognition.

[0022] Preferably, the calculation module calculates the driving direction of the vehicle body based on the position data provided by the GPS positioning module and the steering data provided by the gyroscope, and the control module controls the vehicle body to avoid obstacles according to the driving direction and the position information of the obstacles.

[0023] Preferably, the Internet of Things module is connected to the cloud platform through the Wi-Fi protocol to upload real-time position information and obstacle detection results in real time. The cloud platform is used to display the blind path status and generate a blind path obstacle distribution map.

[0024] Preferably, the calculation module further includes a feedback module for generating a feedback signal according to the environmental information detected by the ultrasonic sensor, the camera and the GPS module. The feedback signal is used to adjust the driving path and speed of the vehicle body.

[0025] The present invention provides a blind path obstacle detection system, which has the following beneficial effects:

[0026] 1. By adopting a solution that combines ultrasonic sensors with image processing technology, the present invention achieves the technical effect of real-time detecting obstacles on the blind path and accurately identifying them. Therefore, through the collaborative work of multiple sensors, the accuracy of obstacle detection is greatly improved, and the problem of missed detection or misjudgment of blind path obstacles caused by a single detection method is avoided.

[0027] 2. By performing time series analysis on image data through the LSTM algorithm, the present invention can accurately predict the movement trajectory of dynamic obstacles, avoid obstacles in advance, and ensure that the vehicle body can drive safely and smoothly in a complex and dynamic blind path environment, greatly improving the obstacle avoidance efficiency.

[0028] 3. By adopting a path planning scheme that combines the Di jkstra algorithm and the ant colony algorithm, the present invention realizes a more intelligent and optimized path selection. Therefore, in terms of path optimization, it can dynamically adjust the path according to the real-time obstacle detection results and environmental changes, avoiding problems such as repeated driving routes or inability to avoid obstacles caused by inflexible path planning.

[0029] 4. Through the Internet of Things module, the present invention can transmit position information and obstacle detection data to the user terminal in real time, improving the monitorability and real-time performance of the system. Compared with traditional blind path navigation systems, the Internet of Things module of the present invention enables users to always master the position and environmental conditions of the vehicle body, ensuring that users can timely respond to possible blind path obstacle situations, and further improving the safety guarantee for blind path users. Description of the Drawings

[0030] Figure 1 is a structural schematic diagram of the present invention;

[0031] Figure 2 is a schematic diagram of the system architecture of the present invention

[0032] Among them, 1. vehicle body; 2. power module; 3. detection module; 4. acquisition module. Specific implementation manners

[0033] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the specification 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 making creative efforts shall fall within the protection scope of the present invention.

[0034] For a better understanding of the present invention, the above content will be described in detail below in conjunction with specific embodiments.

[0035] Please refer to the attached Figure 1 and the attached Figure 2 , an embodiment of the present invention provides a blind path obstacle detection system, including:

[0036] A vehicle body, inside which a detection module, an acquisition module, a GPS positioning module, an auxiliary module, a calculation module, a control module, an Internet of Things module, and a power module are installed;

[0037] The detection module includes an ultrasonic sensor for detecting whether there are obstacles on the blind path;

[0038] In this embodiment, the detection module includes an ultrasonic sensor for detecting whether there are obstacles on the blind path. This ultrasonic sensor cooperates with other modules (such as a calculation module, a control module, etc.) on the vehicle body to realize real-time obstacle monitoring and detection of the blind path.

[0039] Generally, the blind path is a dedicated road for blind people and other visually impaired people to walk. Therefore, it is crucial to keep the blind path unobstructed to ensure their safety. In order to realize real-time monitoring of blind path obstacles, an ultrasonic sensor is used as the detection device in this embodiment. The ultrasonic sensor can detect the distance to the obstacle by emitting ultrasonic waves and receiving the echo signal, thereby realizing the detection of the obstacle.

[0040] Specifically, the ultrasonic sensor can be installed in the front or side of the vehicle body. By emitting ultrasonic waves and receiving the reflected waves, it measures the distance to the obstacle. Based on the measured distance information, the ultrasonic sensor can determine whether there is an obstacle in front, as well as the distance and size of the obstacle. If an obstacle is detected, the ultrasonic sensor will transmit the detected distance information to the calculation module, and the calculation module will perform further obstacle recognition and path planning based on this data.

[0041] In a possible implementation, the measurement range of the ultrasonic sensor can be set to a certain safe distance, such as 2 meters to 3 meters. When the distance between the obstacle and the vehicle body is less than the set safe distance, the system will automatically determine that there is an obstacle and perform corresponding obstacle avoidance processing.

[0042] As an option, the ultrasonic sensor can also be used in conjunction with the camera on the vehicle body. When the ultrasonic sensor detects an obstacle, the system adjusts the angle of the camera by controlling the pan-tilt head to further confirm the type and position of the obstacle. For example, if the ultrasonic sensor detects an obstacle in front, the calculation module can instruct the pan-tilt head to adjust the angle of the camera so that the camera can take a picture of the obstacle, and then use image processing algorithms such as YOLOV8 to analyze the image taken by the camera, identify the obstacle and adjust the motion strategy according to the type of the obstacle.

[0043] Specifically, the output signal of the ultrasonic sensor will be transmitted to the calculation module. The calculation module, by receiving and processing these signals and combining the input information of other sensors (such as the steering angle provided by the gyroscope and the position information provided by the GPS), determines whether there is an obstacle on the blind path and its position. Based on this, the calculation module can also perform path planning and generate a motion control strategy to guide the vehicle body to avoid obstacles or change the driving path.

[0044] In some embodiments, the ultrasonic sensor uses multiple measurement methods to enhance the accuracy of obstacle detection. For example, multiple ultrasonic sensors can be configured to cover the obstacle detection areas in different directions of the vehicle body, and by comprehensively judging the signals output by multiple sensors, the reliability and accuracy of obstacle detection are improved.

[0045] When the ultrasonic sensor detects an obstacle, the calculation module will adjust the driving path and speed of the vehicle body according to the detected distance information. Depending on different obstacle situations, the vehicle body may take measures such as decelerating, stopping, or detouring to avoid obstacles. The collaborative effect of the ultrasonic sensor, the calculation module, and the control module can ensure that the blind path obstacle detection system can respond to the obstacles on the blind path in real time and prevent accidents from occurring to blind path users.

[0046] Therefore, the ultrasonic sensor plays a crucial role in obstacle detection in the vehicle body. Through effective cooperation with other system modules such as the calculation module and the control module, the ultrasonic sensor can detect and judge obstacles on the blind path in real time and provide necessary data support to ensure that the system can perform path planning and obstacle avoidance control according to the actual situation.

[0047] Moreover, in a further embodiment, the detection performance of the ultrasonic sensor can also be optimized according to the actual use environment. For example, in a complex environment, the system can adopt a combination of multiple sensors and various sensing technologies to improve the detection accuracy and ensure the reliable operation of the blind path detection system in different environments.

[0048] The acquisition module includes a camera, which is connected to the vehicle body through a pan-tilt head. The camera is used to collect images of the blind path and obstacles.

[0049] The GPS positioning module is used to obtain the real-time position.

[0050] The auxiliary module includes a gyroscope, which is used to detect the steering angle of the vehicle body.

[0051] In this embodiment, the acquisition module includes a camera, which is connected to the vehicle body through a pan-tilt head. The camera is used to collect images of the blind path and obstacles. The GPS positioning module is used to obtain the real-time position of the vehicle body. The auxiliary module includes a gyroscope, which is used to detect the steering angle of the vehicle body. The cooperation of these modules with other system modules ensures that the blind path obstacle detection system can perform accurate obstacle recognition, path planning, and obstacle avoidance control.

[0052] Specifically, the combination of the camera and the pan-tilt head enables the system to flexibly adjust the angle of the camera to better collect real-time images of the blind path and the surrounding environment. In this embodiment, the cooperation of the camera and the pan-tilt head provides real-time visual information of the environment for the vehicle. These image data can be used for tasks such as obstacle recognition and blind path break detection, providing important inputs for the calculation module.

[0053] Specifically, the camera is connected to the vehicle body through a pan-tilt head and can adjust the angle according to requirements, such as rotating up and down and left and right, to more flexibly collect images of the environment above and in front of the blind path. When the ultrasonic sensor detects an obstacle ahead, the calculation module controls the pan-tilt head to adjust the camera angle according to the feedback of the sensor, so that the camera can accurately capture the obstacle. After the image data is processed, the type of the obstacle can be identified through visual algorithms such as YOLOV8, and corresponding path planning and obstacle avoidance control can be performed according to the type of the obstacle.

[0054] As an option, the camera can also be equipped with high resolution and night vision capabilities to ensure that the system can work effectively in low light environments. In some embodiments, multiple cameras can also be installed as needed, covering different orientations of the vehicle body respectively, to ensure all-round environmental monitoring. The data of multiple cameras can be combined to further improve the accuracy of detecting obstacles on the blind path.

[0055] The GPS positioning module is used to obtain the real-time position data of the vehicle body. Generally, the GPS positioning module calculates the current position of the vehicle body by receiving satellite signals and transmits it to the user terminal in real time via the Internet of Things module. Combining the real-time position data, the system can monitor the position of the vehicle body on the blind path in real time and provide a basis for path planning. Specifically, the position information of the GPS module can be used in coordination with other data in the calculation module (such as the angle information provided by the gyroscope, the image data of the camera, etc.) to ensure that the vehicle body travels precisely on the blind path.

[0056] In a possible implementation, the accuracy of the GPS positioning module can be adjusted according to requirements. For example, differential GPS technology can be used to improve the positioning accuracy. For systems used in urban environments or indoor environments, other sensors (such as inertial navigation systems) can be combined to enhance the stability and accuracy of positioning.

[0057] The gyroscope in the auxiliary module is used to detect the steering angle of the vehicle body and provide real-time angle feedback. This is crucial for the motion control of the vehicle body. Especially when performing path planning and obstacle avoidance operations, the angle information provided by the gyroscope can help the calculation module determine the current driving direction of the vehicle body and perform precise motion control based on this information. For example, when the system detects an obstacle, the calculation module adjusts the steering angle of the vehicle body according to the angle information provided by the gyroscope to avoid collision.

[0058] Specifically, the gyroscope can output the steering angle of the vehicle body in real time, and the calculation module adjusts the driving path of the vehicle body according to this angle information. When the vehicle body travels along the blind path, the system will use the data of the gyroscope to maintain the stability of the vehicle body and ensure its precise travel. For example, when performing path planning, the calculation module will combine the GPS positioning information and the steering data provided by the gyroscope to calculate the current driving trajectory of the vehicle body and adjust the motion direction of the vehicle body in real time to avoid obstacles.

[0059] As an option, the data of the gyroscope can also be fused with the data of other sensors to improve the stability and accuracy of the system. In some embodiments, a multi-axis gyroscope can be used to monitor the motion state of the vehicle body in multiple directions in real time to ensure that the system can control the motion of the vehicle body more precisely.

[0060] Therefore, through the close cooperation of the acquisition module, GPS positioning module, and auxiliary module, the system provides precise monitoring and control of the blind path. The acquisition module provides real-time image data, the GPS positioning module ensures the accuracy of the vehicle body position, and the auxiliary module provides steering angle feedback, providing comprehensive inputs for the calculation module to help achieve path planning and obstacle avoidance functions. Through the effective cooperation of these modules, this embodiment can achieve precise obstacle detection and obstacle avoidance control in a complex blind path environment, ensuring the smoothness and safety of the blind path.

[0061] The calculation module is connected to the ultrasonic sensor, camera, GPS positioning module, and gyroscope. The calculation module performs obstacle detection, path planning, and motion control based on the distance data provided by the ultrasonic sensor, the image data provided by the camera, the position data provided by the GPS positioning module, and the steering angle provided by the gyroscope.

[0062] In this embodiment, the calculation module constructs a comprehensive obstacle detection, path planning, and motion control system by accessing the ultrasonic sensor, camera, GPS positioning module, and gyroscope. This system can detect obstacles in the surrounding environment in real time and perform path planning and vehicle body control based on real-time data to achieve autonomous navigation on the blind path. Specifically, the calculation module performs functions such as obstacle detection, path planning, obstacle avoidance control, and blind path status monitoring based on the input data of each sensor to ensure that the system can adapt to various complex environmental conditions.

[0063] Image Processing and Obstacle Recognition

[0064] Specifically, the calculation module includes an image processing unit. The image processing unit performs blind path obstacle recognition on the image data provided by the camera based on the YOLOV8 algorithm. The YOLOV8 algorithm uses deep learning technology to quickly detect obstacles in the image and classify them according to their categories. The main goal of this algorithm is to perform real-time positioning and recognition of objects in the image and, combined with their category information, adjust the motion control strategy of the control module. Specifically, YOLOV8 can classify different types of obstacles, such as walls, roadblocks, pedestrians, etc., and adjust the motion strategy according to the type of obstacle, such as avoidance or detour.

[0065] In some embodiments, the input of the YOLOV8 algorithm is the image data collected by the camera, and the output is the position and category information of each obstacle in the image. The YOLOV8 algorithm processes the image through a convolutional neural network (CNN) to achieve fast real-time detection. The position of each obstacle in the image is framed and classified according to a preset category. The control module adjusts the motion trajectory of the vehicle body based on this information. For example, when a stationary obstacle is detected, the control module will adjust the speed and direction of the vehicle body to ensure that the vehicle body can safely avoid the obstacle.

[0066] Dynamic obstacle prediction

[0067] In some embodiments, the calculation module performs time series analysis on the image data based on the LSTM (Long Short-Term Memory) algorithm to identify and predict the motion trajectory of dynamic obstacles. The LSTM algorithm can effectively process time series data and predict the future position of the obstacle by analyzing past frame images. The input of the LSTM model is the feature information of consecutive frame images, and the output is the prediction of the motion trajectory of the obstacle. This process can be used for dynamic obstacle avoidance to ensure that the vehicle body can react in advance and adjust the path when encountering a moving obstacle.

[0068] Specifically, the calculation module performs time series analysis on the image data based on the LSTM (Long Short-Term Memory) algorithm to identify and predict the motion trajectory of dynamic obstacles. LSTM can effectively capture the long-term dependencies in time series data and use historical frame image data to predict the future motion trajectory of the obstacle.

[0069] The general structure of the LSTM model includes a forget gate, an input gate, and an output gate, which process the input and output of the image data. During the prediction process, LSTM calculates the state update of the obstacle through the following formula:

[0070] Forget gate:

[0071] f t = σ(W f · [h t-1 , x t + b f );

[0072] Where f t is the output of the forget gate; W f is the weight matrix of the forget gate; h t-1 is the hidden state at the previous moment t-1; x t is the input data at the current moment; b f is the bias term of the forget gate; σ is the sigmoid activation function.

[0073] Input gate:

[0074] i t = σ(W i · [h t-1 , x t + b i );

[0075] Where i t is the output of the input gate; W i is the weight matrix of the input gate; b i is the bias term of the input gate.

[0076] Candidate memory unit:

[0077]

[0078] Among them, is the candidate memory unit; W C is the weight matrix; b C is the candidate memory unit bias term; tanh is the hyperbolic tangent activation function.

[0079] Updated memory unit:

[0080]

[0081] Among them, C t is the memory unit state at the current moment; C t-1 is the memory unit state at the previous moment.

[0082] Output gate:

[0083] o t = σ(W o · [h t-1 , x t + b o );

[0084] Among them, o t is the output of the output gate; W o is the weight matrix of the output gate; b o is the output gate bias term.

[0085] Hidden state:

[0086] h t = o t · tanh(C t );

[0087] Among them, h t is the hidden state at the current moment, and tanh is the hyperbolic tangent activation function.

[0088] Through the above LSTM algorithm, the calculation module can analyze the image data in the past period of time, identify and predict the movement trajectory of dynamic obstacles, and provide a basis for subsequent path planning.

[0089] Path planning and shortest path calculation

[0090] The calculation module generates a path graph of the blind path based on graph theory algorithms and uses the Dijkstra algorithm to calculate the shortest path from the current position information to the target position. The Dijkstra algorithm calculates each node in the path graph one by one to find the shortest path from the starting point to the target point. Specifically, each node in the path graph represents a position in the blind path, the edge represents the connectivity between two nodes, and the weight of the edge represents the degree of influence of the obstacle.

[0091] In some embodiments, the basic formula of the Dijkstra algorithm is as follows:

[0092] Path update:

[0093] d(v) = min(d(v), d(u) + w(u, v));

[0094] Where d(v) is the shortest path distance from node v to the starting point; w(u, v) is the weight of edge (u, v) (i.e., the distance between two points or the degree of influence of the obstacle).

[0095] Shortest path calculation:

[0096] The core of the Dijkstra algorithm is to continuously select the node with the shortest distance from the current starting point and update the shortest distance of adjacent nodes. Eventually, the shortest path from the starting point to each node is calculated.

[0097]

[0098] Where N(v) is the set of neighbor nodes of node v.

[0099] Path optimization and ant colony algorithm

[0100] The calculation module also performs path optimization based on the ant colony algorithm. The ant colony algorithm simulates the process of ants foraging and uses pheromones to update path information. Whenever a path is selected, the pheromone concentration on that path increases, and other ants will choose the path with a higher pheromone concentration, thus achieving the purpose of optimizing the path.

[0101] Specifically, the key formula of the ant colony algorithm is as follows:

[0102] Pheromone update:

[0103] In the ant colony algorithm, the pheromone concentration on the path is updated according to the choice of ants. The update formula for pheromones is:

[0104] τ ij (t + 1) = (1 - ρ)τ ij (t) + Δτ ij ;

[0105] Where τij τ(t) is the pheromone concentration of path (i, j) at time t; τ ij τ(t + 1) is the pheromone concentration of path (i, j) at time t + 1; ρ is the pheromone evaporation coefficient; Δτ ij is the pheromone increment of path (i, j) at time t.

[0106] Pheromone increment:

[0107] The pheromone increment Δτ ij is adjusted according to the quality of path selection (such as path length or the influence of obstacles). Generally, paths with shorter lengths or less influence from obstacles will receive more pheromone increments.

[0108]

[0109] where Q is a constant; L ij is the length (or obstacle influence degree) of path (i, j). The higher the quality of the path, the greater the pheromone increment.

[0110] Path selection probability:

[0111] In the ant colony algorithm, the probability of an ant choosing a path is proportional to the pheromone concentration. The higher the pheromone concentration on the path, the greater the probability of the ant choosing it.

[0112] Subsequently, the control module adjusts the movement trajectory of the vehicle body according to the optimized path to ensure that the vehicle body can avoid obstacles and select the optimal path

[0113] Blind path break detection and Internet of Things communication

[0114] In some embodiments, the calculation module determines whether the blind path has broken by analyzing the changes between the image data provided by the camera and the historical image data. When there are significant differences between the image captured by the camera and the historical image data, the calculation module determines that the blind path may have broken. At this time, the calculation module sends the real-time position information to the user terminal through the Internet of Things module, notifies the user of the status of the blind path, and reminds the user to take necessary measures.

[0115] Obstacle avoidance control

[0116] The calculation module controls the pan-tilt to adjust the angle of the camera according to the obstacle distance data provided by the ultrasonic sensor for obstacle recognition. When the ultrasonic sensor detects an obstacle, the calculation module automatically adjusts the angle of the camera so that it can cover the area where the obstacle is located to ensure accurate identification of the position and type of the obstacle.

[0117] In addition, the calculation module also calculates the driving direction of the vehicle body based on the position data provided by the GPS positioning module and the steering data provided by the gyroscope. Based on the current driving direction and the position information of the obstacles, the control module can adjust the movement trajectory of the vehicle body in real time to ensure that the vehicle body can avoid obstacles and drive smoothly.

[0118] The control module is used to output a motion control strategy according to path planning and motion control;

[0119] The power module is used to drive the vehicle body to move according to the motion control strategy;

[0120] The Internet of Things module is used to transmit the real-time position and obstacle detection information to the user terminal.

[0121] In this embodiment, the control module, the power module, and the Internet of Things module provide precise motion control, path planning, and real-time information transmission functions for the blind path navigation system. The collaborative work of the above modules will ensure that the vehicle body can perform autonomous navigation efficiently and safely in the blind path environment. Specifically, the control module is responsible for outputting control instructions according to path planning and motion control, the power module drives the vehicle body to move according to the control strategy, and the Internet of Things module transmits the position information and obstacle detection data to the user terminal in real time. Through the close cooperation of these three modules, the system can ensure the accuracy, timeliness, and safety of navigation in a complex environment.

[0122] Control module

[0123] The core function of the control module is to output a motion control strategy according to path planning and obstacle detection results, and adjust the driving trajectory of the vehicle body. The path planning module first generates a path map of the blind path based on the surrounding environment and calculates the shortest path from the current position to the target position. The Dijkstra algorithm is used in the calculation process. This algorithm finds the optimal driving path through the shortest path search in graph theory.

[0124] In practical applications, the control module will make real-time path adjustments according to the calculated shortest path in combination with obstacle information. For example, when an obstacle is detected, the control module will process real-time image processing data, and the image information processed by the YOLOV8 algorithm is used to identify the type and position of the obstacle, so as to adjust the motion control strategy to avoid the obstacle.

[0125] Moreover, based on the YOLOv8 instance segmentation technology, accurate identification and tracking of the blind path can be achieved. Specifically, the blind path is accurately segmented by a deep learning model, and a mask image of the blind path area is obtained. This mask image can provide a binary representation of the blind path area, enabling subsequent blind path tracking tasks to be precisely based on the contour of the blind path. By analyzing the pixel points in the mask image, we can extract the center line of the blind path, thus completing the automatic tracking of the blind path. Especially in the case where the colors of the blind path and the road are similar or have the same color, the system can still effectively distinguish and track through the powerful segmentation ability of YOLOv8.

[0126] Specifically, by training a large number of images of blind paths and the surrounding roads, the YOLOv8 model has learned to distinguish the subtle differences between the blind path and the surrounding environment. Even when the colors of the blind path and the road are similar, it can still accurately identify the blind path area. This ability benefits from the high-precision performance of YOLOv8 in instance segmentation tasks. It can not only identify the boundaries of objects but also perform pixel-level precise segmentation in complex environments. The instance segmentation network of YOLOv8 can segment out a clear blind path area in the image regardless of whether the colors of the blind path and the road are similar.

[0127] In practical applications, the system performs instance segmentation on the input image through YOLOv8 to obtain a mask image containing the blind path area. At this time, each pixel point of the blind path will be assigned a specific label, indicating that the pixel belongs to the blind path area. In this way, the system can extract the blind path area from the complex background, avoiding misjudgment in the case of a small color difference between the blind path and the background.

[0128] After the image segmentation is completed, the system analyzes the pixel points of the mask image to identify the pixel distribution within the blind path area. Especially for the tracking of the blind path, the system needs to find a center line within the segmented blind path area, and this line is the tracking path of the blind path. To achieve this goal, this embodiment uses an image processing algorithm to accurately extract the contour of the blind path in the mask image, and determines the traveling route of the blind path by calculating the geometric center of the blind path area in the image.

[0129] Generally, the center line of the blind path area will show a certain curved or straight shape. The system first calculates the pixel distribution of the blind path area through an image analysis algorithm and fits the center line based on the geometric characteristics of the blind path area. To further improve the fitting accuracy, the system can adopt an algorithm based on curve fitting, such as the least squares method, to accurately fit the pixel points of the blind path area and ensure the accuracy of the tracking path.

[0130] As an option, if the blind path presents a relatively complex shape, the system can dynamically adjust the parameters of the fitting algorithm and adopt a more complex fitting method to process the blind path area with complex curves. Through this process, the system can find a smooth and highly accurate center line on the blind path, which serves as the basis for subsequent blind path tracing and obstacle avoidance.

[0131] In addition, in some embodiments, the system further optimizes the segmentation effect of the blind path based on the pixel point distribution in the mask map and combines the texture information of the image. For example, technologies such as image gradient and edge detection are used to enhance the boundary distinction between the blind path and the surrounding environment, ensuring that the edge between the blind path and the road is clearly distinguishable. Through these technical means, the system can further improve the accuracy of blind path segmentation and avoid misjudgment caused by environmental interference.

[0132] Specifically, when processing the mask map, the system first performs a binarization operation to clearly separate the blind path area from the background area. Then, the contour information of the blind path is extracted through an image contour detection algorithm (such as Canny edge detection). Next, a path planning algorithm is used to determine the midline position of the blind path, and the optimal path from the current vehicle position to the target position is calculated through a shortest path algorithm such as Dijkstra's algorithm. The system adjusts the driving direction of the vehicle according to the calculated path to ensure that the vehicle can drive within the blind path.

[0133] In a possible implementation, the system also combines the real-time detection ability of YOLOv8 to update the vehicle's path in real time during the blind path recognition process. Whenever the blind path changes (such as the appearance of new obstacles or road breaks), the system re-segments the image through YOLOv8, updates the mask map, and re-plans the path. This function can ensure the real-time performance and accuracy of the system in a dynamic environment.

[0134] To ensure stability and reliability in complex environments, this embodiment also adopts multi-sensor data fusion technology. The system not only relies on the image segmentation results of YOLOv8 but also combines data from sensors such as ultrasonic sensors, gyroscopes, and GPS positioning. Through data fusion algorithms, the accuracy of blind path recognition and tracing is further improved. Especially when the color of the blind path is similar to the surrounding environment, the fusion of sensor data can effectively make up for the deficiencies of image segmentation and enhance the robustness of the system.

[0135] Generally speaking, through the instance segmentation technology of YOLOv8 in this embodiment, combined with pixel point analysis and path planning algorithms for the blind path area, the goal of precise tracing on the blind path is successfully achieved. Through this technical solution, the system can operate stably in various complex environments, ensuring the safety and convenience of blind path users.

[0136] Based on obstacle detection, the control module outputs corresponding obstacle avoidance control instructions according to the image recognition results, and these control instructions directly affect the execution of the power module. The control module also adjusts the speed, direction, and motion mode of the vehicle body in real time to ensure that the vehicle body can successfully avoid obstacles and reach the target position along the shortest path.

[0137] Power module

[0138] The power module is the execution unit of the system, and its main task is to drive the vehicle body to achieve autonomous movement according to the motion control strategy output by the control module. The instructions output by the control module include the speed, direction, and steering angle of the vehicle body, and these instructions are converted into actual mechanical actions through the execution system of the power module. Specifically, the power module adjusts the traveling trajectory of the vehicle body according to the motion instructions transmitted by the control module to maintain the stable driving of the vehicle body.

[0139] The actuators in the power module usually include motors, servos, etc., which can be adjusted according to the signals sent by the control module. For the situation that requires obstacle avoidance, the power module can quickly respond and make corresponding adjustments according to the motion strategy transmitted by the control module, such as deceleration, steering, or stopping, to ensure the safe movement of the vehicle body.

[0140] In the design of the power module, the control algorithm simulates the vehicle body behavior according to the kinematic model to ensure that the vehicle body movement conforms to the physical laws and avoids abnormal behaviors. This simulation control can generate the optimal motion trajectory under physical constraint conditions according to the current position and target position of the vehicle. The power module needs to be able to quickly respond to these control signals for precise vehicle body control.

[0141] Internet of Things module

[0142] The main function of the Internet of Things module is to transmit the real-time position information of the vehicle body and the obstacle detection data to the user side. Through the GPS positioning module, the Internet of Things module can obtain the current position of the vehicle body and upload this position data to the user side in real time through wireless communication technologies (such as Wi-Fi, 4G / 5G, etc.). The user side can master the driving state of the vehicle body according to these real-time data and conduct remote monitoring.

[0143] In addition, the Internet of Things module can also transmit information about the surrounding environment to the user side through the real-time obstacle detection data to ensure that the user can timely understand the state of the blind path and the possible obstacles. These information includes but is not limited to the type, position of the obstacles and their possible impacts on path planning. The design of the Internet of Things module adopts a low-latency and highly reliable communication protocol to ensure the accuracy and timeliness of real-time data transmission.

[0144] In some embodiments, the Internet of Things module can aggregate and analyze data through a cloud platform to provide feedback for system optimization. For example, by analyzing the data received at the user end, the system can learn and adjust control strategies, thereby continuously improving the obstacle recognition, path planning, and obstacle avoidance capabilities.

[0145] Therefore, through the collaborative work of the control module, the power module, and the Internet of Things module, it is jointly ensured that the system can achieve precise navigation and path planning in a complex environment. The control module generates reasonable control strategies based on the path planning algorithm and the obstacle detection results, the power module drives the vehicle body to safely travel according to these strategies, and the Internet of Things module ensures that the system can communicate with the user end in real time, providing necessary position information and environmental data. This technical solution can fully handle the complex obstacles and dynamic changes in the blind path environment, ensuring the efficient and safe operation of the blind path navigation system.

[0146] 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. A blind path obstacle detection system, characterized in that, Including: A vehicle body (1) with a detection module (3), a collection module (4), a GPS positioning module, an auxiliary module, a calculation module, a control module, an Internet of Things module, and a power module (2) installed inside; The detection module (3) includes ultrasonic sensors for detecting whether there are obstacles on the blind path; The collection module (4) includes a camera and is connected to the vehicle body (1) through a pan-tilt head. The camera is used to collect images of the blind path and obstacles; The GPS positioning module is used to obtain the real-time position; The auxiliary module includes a gyroscope for detecting the steering angle of the vehicle body (1); The calculation module is connected to the ultrasonic sensors, the camera, the GPS positioning module, and the gyroscope. The calculation module performs obstacle detection, path planning, and motion control based on the distance data provided by the ultrasonic sensors, the image data provided by the camera, the position data provided by the GPS positioning module, and the steering angle provided by the gyroscope; The control module is used to output a motion control strategy based on the path planning and motion control; The power module (2) is used to drive the vehicle body (1) to move according to the motion control strategy; The Internet of Things module is used to transmit the real-time position and obstacle detection information to the user terminal.

2. The blind path obstacle detection system according to claim 1, wherein, The calculation module includes an image processing unit. The image processing unit performs blind path obstacle recognition based on the YOLOV8 algorithm. The YOLOV8 algorithm is used to identify the obstacles in the image, classify them according to their categories, and adjust the motion control strategy of the control module according to the categories.

3. The blind path obstacle detection system according to claim 2, characterized in that, The calculation module performs time series analysis on the image data based on the LSTM algorithm to identify and predict the motion trajectories of dynamic obstacles.

4. The blind path obstacle detection system according to claim 2, characterized in that, The calculation module generates a path map of the blind path based on the graph theory algorithm and uses the Dijkstra algorithm to calculate the shortest path from the current position information to the target position. The control module adjusts the motion trajectory of the vehicle body 1 according to the shortest path.

5. The blind path obstacle detection system according to claim 2, wherein, The calculation module performs path optimization based on the ant colony algorithm. The ant colony algorithm updates the pheromone of the path according to the influence degree of obstacles on each path. The control unit adjusts the motion trajectory of the vehicle body 1 according to the optimized path.

6. The blind path obstacle detection system according to claim 1, wherein, When the calculation module detects a break in the blind path, it judges whether the blind path has broken by analyzing the changes between the image data provided by the camera and the historical image data, and sends the real-time position information to the user terminal through the Internet of Things module when a break occurs.

7. The blind path obstacle detection system according to claim 1, characterized in that, When the ultrasonic sensors of the calculation module detect an obstacle, it controls the pan-tilt head to adjust the angle of the camera for obstacle recognition.

8. A blind path obstacle detection system according to claim 1, characterized in that, The calculation module calculates the driving direction of the vehicle body (1) based on the position data provided by the GPS positioning module and the steering data provided by the gyroscope. The control module controls the vehicle body (1) to avoid obstacles according to the driving direction and the position information of the obstacles.

9. The blind path obstacle detection system according to claim 1, characterized in that, The Internet of Things module is connected to the cloud platform through the Wi-Fi protocol, and uploads the real-time position information and obstacle detection results in real time. The cloud platform is used to display the blind path status and generate a blind path obstacle distribution map.

10. A blind path obstacle detection system according to claim 1, characterized in that, The computing module further includes a feedback module for generating a feedback signal based on the environmental information detected by the ultrasonic sensor, camera, and GPS module, and the feedback signal is used to adjust the driving path and speed of the vehicle body 1.

Images