Target detection method for wheeled robot based on trajectory optimization
Through database establishment and image processing technology, the detection path of wheeled robots is optimized, which solves the path rewinding problem and improves detection efficiency and safety.
Patent Information
- Application Number
- CN202411229549.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-03
- Publication Date
- 2024-12-06
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing wheeled robot object detection methods are prone to path reversal when multiple objects are detected, affecting the overall work efficiency.
By establishing a database and using depth cameras and thermal imaging cameras to acquire obstacle images and thermal imaging data, image processing and comparison, calculation of the hardness and heat level of the obstacles, optimizing the detection trajectory, avoiding path reversal, and forming a continuous detection path.
It improves the working efficiency of wheeled robot target detection, avoids direct contact between equipment and high-hardness and high-heat obstacles, and improves detection safety.
Smart Images

Figure CN119091418A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of wheeled robot target detection, and in particular to a wheeled robot target detection method based on trajectory optimization. Background Art
[0002] The existing patent (publication number: CN117095250A) discloses "a lightweight target detection method for a wheeled mobile robot, which relates to the field of artificial intelligence and aims to solve the performance limitations of the edge device of the wheeled mobile robot, as well as the problem that the original YOLOV4 network model is too large and its computing power cannot directly bear the YOLOV4 algorithm, affecting the real-time detection of obstacles. The lightweight target detection method for a wheeled mobile robot disclosed in the present invention greatly reduces the number of parameters and the amount of calculation of the algorithm compared to the original YOLOV4 target detection algorithm. The network weight of the original YOLOV4 is 244MB, and the network model weight provided by the present invention is 77MB. Compared with the original model, the memory usage is reduced by 69%, thereby improving the detection efficiency of the wheeled mobile robot in real-time detection of obstacles, ensuring the safety of the wheeled mobile robot during driving, and meeting the goal of real-time detection of obstacles."
[0003] In the process of implementing this application, the inventor discovered that the prior art has the following problems: when the existing method is used, it is usually adopted to detect obstacles while planning the path according to the position of the obstacles to complete the detection process. When this method is used, when multiple targets need to be detected, repeated path planning may cause the detection path to be repeated, affecting the efficiency of the overall detection work. Summary of the invention
[0004] In view of the shortcomings of the prior art, the present invention provides a wheeled robot target detection method based on trajectory optimization, which solves the problem that when the existing target detection method detects multiple targets continuously, the path of travel may be repeated and the overall work efficiency is poor.
[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: a wheeled robot target detection method based on trajectory optimization, specifically comprising the following steps:
[0006] S1. Database establishment
[0007] Get the material pictures of common objects and the hardness data of the corresponding materials at the same time, input the obtained material pictures and hardness data into the database software, establish a database by corresponding the material pictures and hardness data one by one through the database software, divide the hardness into three levels: low, medium and high, assign values to different levels of hardness, and establish triggers for each material at the same time;
[0008] S2. Equipment Setup
[0009] Build a test site, set up a number of obstacles of different sizes and materials in the test site, bring the wheeled robot to the built test site, mount a depth camera and a thermal imaging camera on the wheeled robot, connect the depth camera and the thermal imaging camera to a host data processing device through a wireless connection, and wirelessly connect the host data processing device to the control system of the wheeled robot;
[0010] S3. Image acquisition
[0011] The wheeled robot is adjusted and moved to an angle that can capture as many obstacles as possible. The depth camera is used to collect images of all obstacles within the visible range, and the thermal imaging camera is used to collect thermal imaging images of all obstacles within the visible range.
[0012] S4. Image processing
[0013] The collected image is transmitted to the upper data processing device, and the obstacles in the image are set as targets through the processing device, and the position of the target is determined, and the focal length coefficient corresponding to the target is set, and the image is processed into multiple copies, and the focal length of each image is processed to focus on a single target. After the focusing is completed, the edge of the target is determined, the target is extracted, the outer contour of the target is removed and saved, and the side length of the target is determined according to the extracted outer contour of the target;
[0014] S5. Image Comparison
[0015] The extracted target image is transmitted to the database, and the target image is compared with the material image in the database to determine the material of the corresponding obstacle. Then, the corresponding hardness data is mobilized through the trigger to determine its hardness level. The heat value of the thermal imaging image of each obstacle is collected, and the heat value is divided into three levels: low, medium and high, and the corresponding heat level is recorded;
[0016] S6. Trajectory calculation optimization
[0017] Substitute the measured obstacle side length and the corresponding hardness and heat values into the trajectory calculation formula, calculate the edge trajectory radius of each obstacle when detecting it according to the trajectory calculation formula, delineate the detection trajectory of each obstacle according to the calculation result, select a point on the delineated trajectory as the target detection point, and connect the points to form a target detection trajectory;
[0018] S7. Object Detection
[0019] Transport the assembled and debugged wheeled robot to the site where target detection is required, repeat the process from S1 to S6, plan the target detection trajectory of the site, and the wheeled robot can move along the designated trajectory, move to the corresponding target detection point in turn and stop, perform target detection, and complete the detection operation of all targets.
[0020] Preferably, in S1, the material hardness is assigned a value of 1 for low hardness, 1.5 for medium hardness, and 2 for high hardness.
[0021] Preferably, in S2, the wireless connection method may be Bluetooth connection or wireless network connection.
[0022] Preferably, in S3, after the acquisition is completed, the position of the wheeled robot is adjusted and multiple shots are taken to ensure that images and thermal imaging images of all obstacles in the venue are acquired.
[0023] Preferably, in S5, the calorie value is assigned as 1 for low calorie, 1.5 for medium calorie, and 2 for high calorie.
[0024] Preferably, in S6, the trajectory calculation formula is:
[0025] Preferably, in the trajectory calculation formula, R is the edge trajectory radius, F 硬 Assign a value to the hardness, F 热 is the heat value, and L is the length of the obstacle.
[0026] Preferably, in S6, when the lengths of the sides of the obstacle are different, the longest side length is substituted into the formula for calculation.
[0027] The present invention provides a wheeled robot target detection method based on trajectory optimization, which has the following features:
[0028] Beneficial effects:
[0029] 1. The present invention provides a wheeled robot target detection method based on trajectory optimization. Compared with the existing target detection method, the target detection method first recognizes the overall obstacle, determines the detection radius of the target according to the hardness and heat of the obstacle, plans the detection path and then performs target detection. This is convenient for the wheeled robot to continuously complete the detection operation on the target that needs to be detected during use, avoiding the path repetition that affects the work efficiency. At the same time, the detection radius is judged by the hardness and heat of the obstacle, avoiding the wheeled robot from being too close to the high-heat and high-hardness obstacle during collection, causing damage to the equipment, thereby improving the safety of the method when performing target detection. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] Figure 1 It is a schematic diagram of the main flow chart of the present invention. DETAILED DESCRIPTION
[0031] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0032] like Figure 1 As shown, an embodiment of the present invention provides a wheeled robot target detection method based on trajectory optimization, which specifically includes the following steps:
[0033] S1. Database establishment
[0034] Get pictures of common object materials, and at the same time, get the hardness data of the corresponding materials, input the obtained material pictures and hardness data into the database software, and establish a database by corresponding the material pictures and hardness data one by one through the database software, so as to facilitate the subsequent image comparison to distinguish materials, and divide the hardness into three levels: low, medium and high, and assign values to different levels of hardness. The material hardness assignment is 1 for low hardness, 1.5 for medium hardness, and 2 for high hardness, so as to facilitate the subsequent use of different detection radii for obstacles of different hardness, and avoid collision and damage when the wheeled robot is in use. At the same time, a trigger is established for each material, so as to facilitate the subsequent retrieval of the material and its related hardness;
[0035] S2. Equipment Setup
[0036] A test site is built, and several obstacles of different sizes and materials are set in the test site. The wheeled robot is brought to the built test site, and a depth camera and a thermal imaging camera are mounted on the wheeled robot. The depth camera and the thermal imaging camera are connected to a host data processing device by wireless connection, and the host data processing device is wirelessly connected to the control system of the wheeled robot. The wireless connection method can be a Bluetooth connection or a wireless network connection, which is conducive to realizing the connection between the devices used in the method, realizing that the data between the devices can be transmitted when the method is operated, and improving the stability of the method when used;
[0037] S3. Image acquisition
[0038] The wheeled robot is adjusted to move to an angle that can capture as many obstacles as possible, and the images of all obstacles within the visible range are collected by the depth camera, and the thermal imaging camera is used to collect thermal imaging images of all obstacles within the visible range. After the collection is completed, the position of the wheeled robot is adjusted and multiple shots are taken to ensure that the images and thermal imaging images of all obstacles in the venue are collected, which is conducive to the device planning all obstacles that need to be detected in the venue;
[0039] S4. Image processing
[0040] The collected image is transmitted to the upper data processing equipment, and the obstacles in the image are set as targets through the processing equipment, and the position of the target is determined, and the focal length coefficient of the corresponding target is set, and the image is processed into multiple copies, and the focal length of each image is processed to focus on a single target. After the focusing is completed, the edge of the target is determined, the target is extracted, and the outer contour of the target is removed and saved. According to the extracted outer contour of the target, the side length of the target is determined, which is convenient for subsequent analysis of a single target, judging its detection path, and realizing the serial connection of the detection process of multiple targets;
[0041] S5. Image Comparison
[0042] The extracted target image is transmitted to the database, and the target image is compared with the material image in the database to determine the material of the corresponding obstacle. Then, the corresponding hardness data is mobilized through the trigger to determine its hardness level. The heat value of the thermal imaging image of each obstacle is collected, and the heat value is divided into three levels: low, medium and high. The heat value is assigned as 1 for low heat, 1.5 for medium heat, and 2 for high heat. The corresponding heat level is recorded to avoid damage to the equipment caused by the high ambient temperature around the target, thereby improving the safety of the overall operation of the method.
[0043] S6. Trajectory calculation optimization
[0044] Substitute the measured obstacle side length and the corresponding hardness and heat values into the trajectory calculation formula. When the obstacle sides are different, take the longest side length value and substitute it into the formula for calculation. According to the trajectory calculation formula, calculate the edge trajectory radius of each obstacle when detecting it. According to the calculation structure, define the detection trajectory of each obstacle. Select a point nearby on the defined trajectory as the target detection point, connect the points, and form a target detection trajectory. The trajectory calculation formula is: In the trajectory calculation formula, R is the edge trajectory radius, F 硬 Assign a value to the hardness, F 热 is the heat value, and L is the length of the obstacle side, so as to facilitate the calculation of the path when the method is actually operated;
[0045] S7. Object Detection
[0046] Transport the assembled and debugged wheeled robot to the site where target detection is required, repeat the process from S1 to S6, plan the target detection trajectory of the site, and the wheeled robot can move along the designated trajectory, move to the corresponding target detection point in turn and stop, perform target detection, and complete the detection operation of all targets, so as to facilitate the application of this method to different sites.
[0047] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.
Claims
1. A wheeled robot target detection method based on trajectory optimization, characterized in that: The specific steps include: S1. Database establishment Get the material pictures of common objects and the hardness data of the corresponding materials at the same time, input the obtained material pictures and hardness data into the database software, establish a database by corresponding the material pictures and hardness data one by one through the database software, divide the hardness into three levels: low, medium and high, assign values to different levels of hardness, and establish triggers for each material at the same time; S2. Equipment Setup Build a test site, set up a number of obstacles of different sizes and materials in the test site, bring the wheeled robot to the built test site, mount a depth camera and a thermal imaging camera on the wheeled robot, connect the depth camera and the thermal imaging camera to a host data processing device through a wireless connection, and wirelessly connect the host data processing device to the control system of the wheeled robot; S3. Image acquisition The wheeled robot is adjusted and moved to an angle that can capture as many obstacles as possible. The depth camera is used to collect images of all obstacles within the visible range, and the thermal imaging camera is used to collect thermal imaging images of all obstacles within the visible range. S4. Image processing The collected image is transmitted to the upper data processing device, and the obstacles in the image are set as targets through the processing device, and the position of the target is determined, and the focal length coefficient corresponding to the target is set, and the image is processed into multiple copies, and the focal length of each image is processed to focus on a single target. After the focusing is completed, the edge of the target is determined, the target is extracted, the outer contour of the target is removed and saved, and the side length of the target is determined according to the extracted outer contour of the target; S5. Image Comparison The extracted target image is transmitted to the database, and the target image is compared with the material image in the database to determine the material of the corresponding obstacle. Then, the corresponding hardness data is mobilized through the trigger to determine its hardness level. The heat value of the thermal imaging image of each obstacle is collected, and the heat value is divided into three levels: low, medium and high, and the corresponding heat level is recorded; S6. Trajectory calculation optimization Substitute the measured obstacle side length and the corresponding hardness and heat values into the trajectory calculation formula, calculate the edge trajectory radius of each obstacle when detecting it according to the trajectory calculation formula, delineate the detection trajectory of each obstacle according to the calculation result, select a point on the delineated trajectory as the target detection point, and connect the points to form a target detection trajectory; S7. Object Detection Transport the assembled and debugged wheeled robot to the site where target detection is required, repeat the process from S1 to S6, plan the target detection trajectory of the site, and the wheeled robot can move along the designated trajectory, move to the corresponding target detection point in turn and stop, perform target detection, and complete the detection operation of all targets.
2. The wheeled robot target detection method based on trajectory optimization according to claim 1, characterized in that: In S1, the material hardness is assigned a value of 1 for low hardness, 1.5 for medium hardness, and 2 for high hardness.
3. The wheeled robot target detection method based on trajectory optimization according to claim 1, characterized in that: In S2, the wireless connection method may be Bluetooth connection or wireless network connection.
4. The wheeled robot target detection method based on trajectory optimization according to claim 1, characterized in that: In S3, after the acquisition is completed, the position of the wheeled robot is adjusted, and multiple shots are taken to ensure that images and thermal imaging images of all obstacles in the venue are acquired.
5. The wheeled robot target detection method based on trajectory optimization according to claim 1, characterized in that: In S5, the calorie value is assigned as 1 for low calorie, 1.5 for medium calorie and 2 for high calorie.
6. The wheeled robot target detection method based on trajectory optimization according to claim 1, characterized in that: In S6, the trajectory calculation formula is:
7. The wheeled robot target detection method based on trajectory optimization according to claim 6, characterized in that: In the trajectory calculation formula, R is the edge trajectory radius, F 硬 Assign a value to the hardness, F 热 is the heat value, and L is the length of the obstacle.
8. The wheeled robot target detection method based on trajectory optimization according to claim 1, characterized in that: In S6, when the lengths of the sides of the obstacle are different, the longest side length is substituted into the formula for calculation.
