Unmanned pedestrian detection method

Abstract

The invention relates to the technical field of pedestrian detection, in particular to an unmanned pedestrian detection method, which comprises the following steps of: matching other data transmission modules within 10 meters in real time; when the data transmission module is successfully matched with other data transmission modules, the first real-time pedestrian detection information is transmitted to other detection systems to generate second real-time pedestrian detection information; performing identical matching on the first real-time pedestrian detection information output by the deep learning models in other detection systems and the second real-time pedestrian detection information to obtain third real-time pedestrian detection information which is not successfully matched; and the other detection systems comprehensively judge the position and the dynamic state of the pedestrian based on the first real-time pedestrian detection information in the other detection systems and the obtained third real-time pedestrian detection information. According to the invention, the shielded vehicle can obtain the information of the shielded pedestrian, so that the situation of missing detection at the sidewalk is avoided, and the safety of unmanned driving is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of pedestrian detection, and in particular to an unmanned driving pedestrian detection method. Background Art

[0002] With the rapid development of the automobile industry and the continuous improvement of people's living conditions, cars have become one of the indispensable means of transportation for people to travel. Autonomous vehicles are a type of intelligent vehicles, which mainly rely on the intelligent driving instrument based on the computer system in the car to achieve the purpose of autonomous driving.

[0003] In unmanned driving technology, pedestrian detection is the focus of unmanned driving technology. Pedestrians are easily obscured by other objects in some cases. For example, at a pedestrian crossing, vehicles in the next lane will obstruct pedestrians, resulting in poor robustness of existing pedestrian detection methods, which can easily lead to missed detections and the risk of "ghost poking their heads out", which will ultimately affect the safety of unmanned driving. Summary of the invention

[0004] In view of this, the purpose of the present invention is to propose an unmanned driving pedestrian detection method to solve the problem that the existing pedestrian detection is prone to missed detection due to occlusion.

[0005] Based on the above purpose, the present invention provides an unmanned driving pedestrian detection method, comprising the following steps:

[0006] S1: Continuously collect data about the vehicle's surrounding environment based on multiple sensors, and obtain the primary data collected by each sensor;

[0007] S2: preprocessing the primary data collected by each sensor to obtain the preprocessed secondary data;

[0008] S3: Based on the data fusion algorithm, different secondary data are fused to obtain tertiary data;

[0009] S4: input the third-level data into the deep learning model for analysis to obtain first real-time pedestrian detection information;

[0010] S5: transmitting the first real-time pedestrian detection information to a data transmission module;

[0011] S6: Real-time matching with other data transmission modules within 10 meters;

[0012] S7: When the data transmission module successfully matches with other data transmission modules, the first real-time pedestrian detection information is transmitted to other detection systems to generate second real-time pedestrian detection information;

[0013] S8: performing identity matching on the first real-time pedestrian detection information output by the deep learning model in the other detection system and the second real-time pedestrian detection information to obtain third real-time pedestrian detection information that has not been successfully matched;

[0014] S9: The other detection systems comprehensively determine the position and dynamics of the pedestrian based on the first real-time pedestrian detection information in the other detection systems and the acquired third real-time pedestrian detection information.

[0015] Optionally, after step S9, the step further includes:

[0016] Based on the position and dynamics of pedestrians, it is determined whether there is potential danger. When it is determined that there is potential danger, the early warning mechanism is triggered in time.

[0017] Optionally, the warning mechanism at least includes controlling the vehicle to slow down and sounding a horn as a warning.

[0018] Optionally, the multiple sensors include cameras, infrared sensors and radars.

[0019] Optionally, the preprocessing at least includes:

[0020] Data cleaning: used to remove invalid data corresponding to noise and outliers;

[0021] Data calibration: ensuring that data between different sensors are consistent in time and space;

[0022] Data synchronization: Synchronize data from different sensors to the same timestamp.

[0023] Optionally, the fusing of different secondary data includes:

[0024] Extract image features such as edges, corners, and textures from camera data;

[0025] Extract characteristics of temperature distribution and heat sources from infrared sensor data;

[0026] Extract range, velocity, and angle features from radar data;

[0027] Based on the feature-level fusion principle, the above features are fused.

[0028] Optionally, the data transmission module is a Bluetooth module.

[0029] When detecting pedestrians through this detection method, the data of the vehicle's surrounding environment is first continuously collected through various sensors on the vehicle to obtain the primary data collected by each sensor, and then the primary data is preprocessed to obtain the preprocessed secondary data. The corresponding secondary data obtained by different sensors are fused based on the data fusion algorithm to obtain the fused tertiary data, and then the tertiary data is input into the deep learning model for analysis to obtain the first real-time pedestrian detection information. The first real-time pedestrian detection information represents the information of unobstructed pedestrians that can be directly detected by the vehicle. The first real-time pedestrian detection information is then transmitted to the data transmission module to match other data transmission modules within 10 meters in real time. When the data transmission module is matched with other data After the transmission module matches successfully, the first real-time pedestrian detection information is transmitted to other detection systems to generate second real-time pedestrian detection information, and the first real-time pedestrian detection information output by the deep learning model in other detection systems is identically matched with the second real-time pedestrian detection information to obtain third real-time pedestrian detection information that has not been successfully matched. The third real-time pedestrian detection information represents the obscured pedestrian information that has not been detected in other detection systems, but the third real-time pedestrian detection information is detected by other vehicles and then passed to the measurement that cannot be detected due to obstruction. Other detection systems comprehensively judge the position and dynamics of pedestrians based on the first real-time pedestrian detection information in other detection systems and the obtained third real-time pedestrian detection information.

[0030] From the above, it can be seen that through the above method, a vehicle can obtain the pedestrian information detected by other vehicles. When a pedestrian is blocked by other vehicles, the blocking vehicle can detect the pedestrian. The pedestrian information detected by the blocking vehicle is transmitted to the blocked vehicle, so that the blocked vehicle can obtain the blocked pedestrian information, thereby avoiding missed detection on the sidewalk and improving the safety of unmanned driving. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0032] Figure 1 The figure is a flow chart of a pedestrian detection method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0033] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0034] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0035] like Figure 1 As shown, a driverless pedestrian detection method includes the following steps:

[0036] S1: Continuously collect data about the vehicle's surrounding environment based on multiple sensors, and obtain the primary data collected by each sensor;

[0037] S2: preprocessing the primary data collected by each sensor to obtain the preprocessed secondary data;

[0038] S3: Based on the data fusion algorithm, different secondary data are fused to obtain tertiary data;

[0039] S4: input the third-level data into the deep learning model for analysis to obtain first real-time pedestrian detection information;

[0040] S5: transmitting the first real-time pedestrian detection information to a data transmission module;

[0041] S6: Real-time matching with other data transmission modules within 10 meters;

[0042] S7: When the data transmission module successfully matches with other data transmission modules, the first real-time pedestrian detection information is transmitted to other detection systems to generate second real-time pedestrian detection information;

[0043] S8: performing identity matching on the first real-time pedestrian detection information output by the deep learning model in the other detection system and the second real-time pedestrian detection information to obtain third real-time pedestrian detection information that has not been successfully matched;

[0044] S9: The other detection systems comprehensively determine the position and dynamics of the pedestrian based on the first real-time pedestrian detection information in the other detection systems and the acquired third real-time pedestrian detection information.

[0045] When detecting pedestrians through this detection method, the data of the vehicle's surrounding environment is first continuously collected through various sensors on the vehicle to obtain the primary data collected by each sensor, and then the primary data is preprocessed to obtain the preprocessed secondary data. The corresponding secondary data obtained by different sensors are fused based on the data fusion algorithm to obtain the fused tertiary data, and then the tertiary data is input into the deep learning model for analysis to obtain the first real-time pedestrian detection information. The first real-time pedestrian detection information represents the information of unobstructed pedestrians that can be directly detected by the vehicle. The first real-time pedestrian detection information is then transmitted to the data transmission module to match other data transmission modules within 10 meters in real time. When the data transmission module is matched with other data After the transmission module matches successfully, the first real-time pedestrian detection information is transmitted to other detection systems to generate second real-time pedestrian detection information, and the first real-time pedestrian detection information output by the deep learning model in other detection systems is identically matched with the second real-time pedestrian detection information to obtain third real-time pedestrian detection information that has not been successfully matched. The third real-time pedestrian detection information represents the obscured pedestrian information that has not been detected in other detection systems, but the third real-time pedestrian detection information is detected by other vehicles and then passed to the measurement that cannot be detected due to obstruction. Other detection systems comprehensively judge the position and dynamics of pedestrians based on the first real-time pedestrian detection information in other detection systems and the obtained third real-time pedestrian detection information.

[0046] From the above, it can be seen that through the above method, a vehicle can obtain the pedestrian information detected by other vehicles. When a pedestrian is blocked by other vehicles, the blocking vehicle can detect the pedestrian. The pedestrian information detected by the blocking vehicle is transmitted to the blocked vehicle, so that the blocked vehicle can obtain the blocked pedestrian information, thereby avoiding missed detection on the sidewalk and improving the safety of unmanned driving.

[0047] In some embodiments, the step S9 further includes:

[0048] Based on the position and dynamics of pedestrians, it is determined whether there is a potential danger. When it is determined that there is a potential danger, the early warning mechanism is triggered in time. Optionally, the early warning mechanism at least includes controlling the vehicle to slow down and sounding the horn for early warning. For example, when it is determined that there is a danger of "ghost poking its head", controlling the vehicle to slow down and sounding the horn for early warning can effectively remind the people in the vehicle and pedestrians on the road, reducing the risk of collision.

[0049] In some embodiments, the multiple sensors include cameras, infrared sensors, and radars. The information obtained by the camera may include roads, pedestrians, other vehicles, traffic signals, road markings, etc. The information obtained by the infrared sensor may include heat sources such as the road surface, weather, and other vehicles. The infrared sensor can help the vehicle measure and track surrounding objects more accurately and enhance perception and decision-making capabilities. Radars include laser radars, millimeter-wave radars, ultrasonic radars, etc. The information obtained may include the distance, speed, and direction of objects. These data are essential for avoiding collisions, maintaining safe vehicle distances, and realizing functions such as adaptive cruise control. Laser radars (especially multi-line laser radars) can obtain depth information within a spatial range, thereby constructing a detailed three-dimensional model of the vehicle's surroundings, which helps unmanned vehicles to have a more comprehensive understanding of the surrounding environment and make more accurate driving decisions.

[0050] In some embodiments, the preprocessing includes at least:

[0051] Data cleaning: used to remove invalid data corresponding to noise and outliers;

[0052] Data calibration: ensuring that data between different sensors are consistent in time and space;

[0053] Data synchronization: Synchronize data from different sensors to the same timestamp.

[0054] After preprocessing, the processed data can be subjected to subsequent fusion processing.

[0055] In some embodiments, the fusing of different secondary data includes:

[0056] Extract image features from camera data, such as edges, corners, and textures;

[0057] Extract features such as temperature distribution and heat sources from infrared sensor data;

[0058] Extract features such as distance, speed, and angle from radar data;

[0059] Based on the feature-level fusion principle, the above features are fused.

[0060] By fusing multiple feature data, the system can detect pedestrians more reliably and reduce false alarms and missed alarms.

[0061] For example, camera and radar fusion: the camera provides rich visual information, such as color, texture, shape, etc., and the radar provides accurate distance and speed information to locate obstacles. The camera and radar data are matched and associated through the fusion algorithm to form a complete perception of obstacles;

[0062] Camera and infrared sensor fusion: Cameras perform well during the day and in good lighting conditions, while infrared sensors provide supplementary information, such as heat source detection, at night or in low light conditions. Fusion algorithms can combine the advantages of both to improve the vehicle's perception capabilities in various lighting conditions;

[0063] Multi-sensor comprehensive fusion: The feature data of multiple sensors such as cameras, infrared sensors, radars, etc. are comprehensively integrated to form a comprehensive and accurate perception of the surrounding environment by leveraging their respective advantages and complementarity.

[0064] In some embodiments, the data transmission module is a Bluetooth module.

[0065] It should be understood by those skilled in the art that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present invention (including the claims) is limited to these examples. Within the scope of the present invention, the technical features in the above embodiments or in different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the above aspects of the present invention, which are not provided in detail for the sake of simplicity.

[0066] The present invention is intended to cover all such substitutions, modifications and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A driverless pedestrian detection method, characterized in that: The following steps are involved: S1: Continuously collect data about the vehicle's surrounding environment based on multiple sensors, and obtain the primary data collected by each sensor; S2: preprocessing the primary data collected by each sensor to obtain the preprocessed secondary data; S3: Based on the data fusion algorithm, different secondary data are fused to obtain tertiary data; S4: input the third-level data into the deep learning model for analysis to obtain first real-time pedestrian detection information; S5: transmitting the first real-time pedestrian detection information to a data transmission module; S6: Real-time matching with other data transmission modules within 10 meters; S7: When the data transmission module successfully matches with other data transmission modules, the first real-time pedestrian detection information is transmitted to other detection systems to generate second real-time pedestrian detection information; S8: performing identity matching on the first real-time pedestrian detection information output by the deep learning model in the other detection system and the second real-time pedestrian detection information to obtain third real-time pedestrian detection information that has not been successfully matched; S9: The other detection systems comprehensively determine the position and dynamics of the pedestrian based on the first real-time pedestrian detection information in the other detection systems and the acquired third real-time pedestrian detection information.

2. The unmanned driving pedestrian detection method according to claim 1, characterized in that: The step S9 further includes: Based on the position and dynamics of pedestrians, it is determined whether there is potential danger. When it is determined that there is potential danger, the early warning mechanism is triggered in time.

3. The unmanned driving pedestrian detection method according to claim 2, characterized in that: The warning mechanism at least includes controlling the vehicle to slow down and sounding the horn to warn.

4. The unmanned driving pedestrian detection method according to claim 1, characterized in that: The various sensors include cameras, infrared sensors and radars.

5. The unmanned driving pedestrian detection method according to claim 4, characterized in that: The pre-processing at least comprises: Data cleaning: used to remove invalid data corresponding to noise and outliers; Data calibration: ensuring that data between different sensors are consistent in time and space; Data synchronization: Synchronize data from different sensors to the same timestamp.

6. The unmanned driving pedestrian detection method according to claim 5, characterized in that: The fusion processing of different secondary data includes: Extract image features such as edges, corners, and textures from camera data; Extract characteristics of temperature distribution and heat sources from infrared sensor data; Extract range, velocity, and angle features from radar data; Based on the feature-level fusion principle, the above features are fused.

7. The unmanned driving pedestrian detection method according to claim 1, characterized in that: The data transmission module is a Bluetooth module.

Images