A behavior driving method applied when a robot returns to a work station

Abstract

The application discloses a behavior driving method applied to a robot returning to a work station, and comprises the following steps: a robot end is taken as a positioning terminal, interacts with the work station, obtains two coordinates of the robot end and the base station, autonomously updates the position of the work station in a map, updates a new return station point in combination with the map, and then the robot starts returning to the station according to the updated work station point; after contacting the work station, the robot immediately obtains an electric level piece contact signal through a relay on the pole piece, at this time, a high-precision ultrasonic function is used to detect the distance of the work station, and the charging level piece compression amount is back calculated, when reaching a preset value, the actual current of two motors at this time is calculated, the motor current and the level piece compression amount are brought into a database for fitting, and whether the robot is normally returned to the position is judged; the robot can be prevented from damaging the work station; the service life of the motor and the charging electrode is prolonged, wear and tear of the drain groove and hole is reduced; standby power consumption of the work station is reduced; and the success rate and reliability of returning to the station are increased.

Description

Technical Field

[0001] This invention relates to the field of intelligent robot technology, specifically a behavior-driven method applied to the return of a robot to its workstation. Background Technology

[0002] With the development of technology, household cleaning equipment has evolved from handheld vacuum cleaners to washer-mop combos, to traction washer-mop combos, and finally to washer-mop robots; commercial cleaning has progressed from manual washer-mop combos to semi-automatic and fully automatic washer-mop combos. Cleaning robots are becoming increasingly independent; current cleaning robots require almost no human intervention to complete 24 / 7 unattended, scheduled cleaning tasks. They can autonomously add water, empty wastewater, recharge, and wake up, automatically completing cleaning tasks. Autonomous return to the charging station, adding water, and emptying wastewater are crucial for truly unattended operation.

[0003] For the autonomous return-to-station function, the current market uses a combination of map navigation, image recognition, and workstation interaction. First, the map guides the robot to the front of the workstation, then the image recognition is applied to the workstation, and then the drive motor slowly enters the station. When the robot touches the charging station, information is exchanged through the contact point, the motor is locked, and the return-to-station function is completed.

[0004] The robot's return to the workstation relies on the initial workstation position marked during mapping (initial point A). If the workstation is moved left or right (the new point B), the robot will return to the position directly in front of the initial workstation (point A), not directly in front of its current position (point B). This will affect the next step of the workstation alignment operation because the robot is no longer on the workstation's central axis. During alignment, an angle will be created, causing the robot to approach the workstation at an angle instead of facing it directly. This results in an oblique force on the protruding water inlet and outlet of the workstation and on the robot's water inlet and outlet holes. This force can cause the workstation to tilt or even be pushed up. The reaction force ultimately acts on the hub motor. If the robot and workstation complete a handshake at this point, locking the hub motor, the load on the hub motor will increase. This increased load will raise the motor temperature, potentially triggering over-temperature protection. Frequent occurrences of this could demagnetize or burn out the motor. When the motor overheats and triggers the overheat protection, the machine will be pushed out of the workstation by the reaction force. At this time, charging and water supply / drainage are in progress, and the water supply and drainage channels cannot be immediately shut off, causing water to splash onto the robot and the charging station's electrode plates, creating a potential hazard. The lifespan of the hub motor and charging electrode plates is significantly reduced, increasing after-sales costs. Furthermore, a malfunction in the motherboard communicating with the robot in the workstation could cause the machine to stubbornly resist the workstation, resulting in damage. These multiple factors reduce the success rate of returning to the station, affecting work efficiency, and any of these occurrences will inevitably impact customer trust in the product.

[0005] Therefore, we propose a behavior-driven approach for robots returning to their workstations to address the aforementioned problems. Summary of the Invention

[0006] The purpose of this invention is to provide a behavior-driven method for robots returning to their workstations, in order to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A behavior-driven method applied to the return of a robot to its workstation includes the following steps:

[0009] Step 1: The workstation acts as a UWB base station, using an array antenna. The robot acts as a positioning terminal, interacting with the workstation to obtain its own and the base station's two-dimensional coordinates. It autonomously updates the workstation's position on the map. When returning to the station, it automatically travels to the workstation's central axis. By using the time difference of electromagnetic waves from any two antennas on the array antenna to the terminal, two points in the plane can be determined. By combining the phase and time difference of the electromagnetic waves received by the other antennas, the specific orientation can be obtained. Finally, a unique point is determined, and the new return station is updated based on the map.

[0010] Step 2: Then, the robot travels to the central axis of the workstation according to the updated workstation and begins to return to the station. Using image recognition and ultrasonic recognition functions, the maximum drive current of the hub motor is reduced when returning to the station, so that it can push the charging plate back to the bottom with an additional force of 5-10N on the basis of normal walking, so as to obtain the maximum working current when returning to the station.

[0011] Step 3: Before contacting the workstation, the robot is guided by an image and map. After contacting the workstation, the robot immediately receives a contact signal from the relay on the electrode plate. At this time, the robot starts to use high-precision ultrasonic function to detect the distance to the workstation and calculates the compression amount of the electrode plate. When the preset value is reached, the robot stops moving forward and calculates the actual current of the two motors. The motor current and the compression amount of the electrode plate are then used to fit the data in the database to determine whether the robot has returned to the station normally.

[0012] As a further aspect of the present invention: the database is a database compiled from a large amount of experimental data in the early stages.

[0013] As a further aspect of the present invention: the image recognition function performs image recognition through an image recognition model, and the training method of the image recognition model includes: acquiring an initial training image, performing image signal processing augmentation on the initial training image to obtain a processed training image; and training the image recognition model based on the processed training image.

[0014] As a further aspect of the present invention, the image signal processing augmentation of the initial training image includes performing at least one of the following processes on the initial training image: RGB domain augmentation, HSV domain augmentation, and YUV domain augmentation.

[0015] As a further aspect of the present invention: the RGB domain augmentation processing includes at least one of color information adjustment, gamma transformation, and random histogram equalization; wherein, the parameters in the color correction matrix and bias matrix used in color information adjustment are obtained by fine-tuning preset parameters based on random variables; the gamma coefficients used in gamma transformation are obtained by fine-tuning preset parameters based on random variables; and random histogram equalization refers to determining whether to perform histogram equalization based on random variables.

[0016] As a further aspect of the present invention: when the RGB domain augmentation process includes at least two of color information adjustment, gamma transformation and random histogram equalization, the color information adjustment is performed before the gamma transformation and the gamma transformation is performed before the random histogram equalization.

[0017] As a further aspect of the present invention: the HSV domain augmentation processing includes saturation adjustment and / or contrast adjustment; wherein, the saturation parameter used in the saturation adjustment is obtained by fine-tuning a preset parameter based on a random variable, and the contrast parameter used in the contrast adjustment is obtained by fine-tuning a preset parameter based on a random variable.

[0018] As a further aspect of the present invention: the YUV domain augmentation processing includes YUV domain noise reduction and / or edge enhancement; wherein, the parameters of the low-pass filter used in YUV domain noise reduction are obtained by fine-tuning preset parameters based on random variables, and the parameters used in edge enhancement are obtained by fine-tuning preset parameters based on random variables.

[0019] As a further aspect of the present invention: the ultrasonic recognition function includes acquiring ultrasonic detection signals during recognition.

[0020] As a further aspect of the present invention: the ultrasonic detection signal is an electrical signal obtained by converting the ultrasonic signal sensed by the ultrasonic sensor into an electrical signal. Before decomposing the ultrasonic detection signal into a specified number of signal components with equal bandwidth, the method further includes: filtering the ultrasonic detection signal to obtain a filtered ultrasonic detection signal; correspondingly, decomposing the ultrasonic detection signal into a specified number of signal components with equal bandwidth means: decomposing the filtered ultrasonic detection signal into a specified number of signal components with equal bandwidth.

[0021] Compared with the prior art, the beneficial effects of the present invention are:

[0022] 1. Prevent damage to the workstation.

[0023] 2. Extend the lifespan of the motor and charging electrodes, and reduce wear and tear on the water inlet / outlet.

[0024] 3. Reduce workstation standby power consumption.

[0025] 4. Increase the success rate and reliability of returning to the station. Attached Figure Description

[0026] Figure 1 This is a flowchart illustrating the behavior-driven method for the robot returning to the workstation in this invention.

[0027] Figure 2 This is a schematic diagram of the two-position positioning principle of the UWB array antenna in this invention.

[0028] Figure 3 This is a schematic diagram of the robot returning to its station in this invention. Detailed Implementation

[0029] In one embodiment, such as Figures 1-3 As shown, a behavior-driven method applied to the return of a robot to its workstation includes the following steps:

[0030] Step 1: The workstation acts as a UWB base station, using an array antenna. The robot acts as a positioning terminal, interacting with the workstation to obtain its own and the base station's two-dimensional coordinates. It autonomously updates the workstation's position on the map. When returning to the station, it automatically travels to the workstation's central axis. By using the time difference of electromagnetic waves from any two antennas on the array antenna to the terminal, two points in the plane can be determined. By combining the phase and time difference of the electromagnetic waves received by the other antennas, the specific orientation can be obtained. Finally, a unique point is determined, and the new return station is updated based on the map.

[0031] Step 2: Then, the robot travels to the central axis of the workstation according to the updated workstation and begins to return to the station. Using image recognition and ultrasonic recognition functions, the maximum drive current of the hub motor is reduced when returning to the station, so that it can push the charging plate back to the bottom with an additional force of +5-10N on the basis of normal walking, so as to obtain the maximum working current (I) when returning to the station.

[0032] Step 3: Before contacting the workstation, the robot is guided by an image and map. After contacting the workstation, the robot immediately receives a contact signal (Dock signal) from the relay on the electrode plate. At this time, the robot starts to use high-precision ultrasonic function to detect the distance to the workstation and calculates the compression amount of the electrode plate. When the preset value (L1) is reached, the robot stops moving forward and calculates the actual current of the two motors. The motor current and the compression amount of the electrode plate are then used to fit the data in the database to determine whether the robot has returned to the workstation normally.

[0033] The database was created based on a large amount of experimental data from the previous stage.

[0034] This invention eliminates the possibility of damaging the workstation; extends the lifespan of the motor and charging electrodes, reduces wear and tear on the filler / drainage channels and holes; reduces the workstation's standby power consumption; and increases the success rate and reliability of returning to the station.

[0035] The image recognition function performs image recognition through an image recognition model. The training method of the image recognition model includes: acquiring an initial training image; performing image signal processing augmentation on the initial training image to obtain a processed training image; and training the image recognition model based on the processed training image.

[0036] Augmentation processing of the initial training image by image signal processing includes: performing at least one of the following processing on the initial training image: RGB domain augmentation processing, HSV domain augmentation processing, and YUV domain augmentation processing;

[0037] RGB domain augmentation processing includes at least one of color information adjustment, gamma transformation, and random histogram equalization; wherein, the parameters in the color correction matrix and bias matrix used in color information adjustment are obtained by fine-tuning preset parameters based on random variables; the gamma coefficients used in gamma transformation are obtained by fine-tuning preset parameters based on random variables; random histogram equalization refers to determining whether to perform histogram equalization based on random variables;

[0038] When RGB domain augmentation processing includes at least two of the following: color information adjustment, gamma transformation, and random histogram equalization, color information adjustment is performed before gamma transformation, and gamma transformation is performed before random histogram equalization.

[0039] HSV domain augmentation processing includes saturation adjustment and / or contrast adjustment; wherein, the saturation parameter used in saturation adjustment is obtained by fine-tuning preset parameters based on random variables, and the contrast parameter used in contrast adjustment is obtained by fine-tuning preset parameters based on random variables.

[0040] YUV domain augmentation processing includes YUV domain noise reduction and / or edge enhancement; wherein, the parameters of the low-pass filter used in YUV domain noise reduction are obtained by fine-tuning preset parameters based on random variables, and the parameters used in edge enhancement are obtained by fine-tuning preset parameters based on random variables.

[0041] The training method of image recognition model can achieve the effect of training images taken by various cameras by performing image signal processing augmentation on the initial training images. Therefore, the image recognition model trained on the augmented images will enhance the generalization of different modules, so that the image recognition model has good performance when performing image recognition on images to be recognized taken by different imaging modules.

[0042] The ultrasonic recognition function includes acquiring an ultrasonic detection signal during recognition. The ultrasonic detection signal is an electrical signal obtained by converting the ultrasonic signal sensed by the ultrasonic sensor into an electrical signal. Before decomposing the ultrasonic detection signal into a specified number of signal components with equal bandwidth, the function also includes filtering the ultrasonic detection signal to obtain a filtered ultrasonic detection signal. Correspondingly, decomposing the ultrasonic detection signal into a specified number of signal components with equal bandwidth means decomposing the filtered ultrasonic detection signal into a specified number of signal components with equal bandwidth.

[0043] By filtering the ultrasonic detection signal, the influence of interference signals such as white noise can be removed, further improving the accuracy of subsequent feature extraction.

[0044] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A behavior-driven method applied to a robot returning to its workstation, characterized in that, Includes the following steps: Step 1: The workstation acts as a UWB base station, using an array antenna. The robot acts as a positioning terminal, interacting with the workstation to obtain its own and the base station's two-dimensional coordinates. It autonomously updates the workstation's position on the map. When returning to the station, it automatically travels to the workstation's central axis. By using the time difference of electromagnetic waves from any two antennas on the array antenna to the terminal, two points in the plane can be determined. By combining the phase and time difference of the electromagnetic waves received by the other antennas, the specific orientation can be obtained. Finally, a unique point is determined, and the new return station is updated based on the map. Step 2: Then, the robot travels to the central axis of the workstation according to the updated workstation and begins to return to the station. Using image recognition and ultrasonic recognition functions, the maximum drive current of the hub motor is reduced when returning to the station, so that it can push the charging plate back to the bottom with an additional force of 5 to 10N on the basis of normal walking, thus obtaining the maximum working current when returning to the station. Step 3: Before contacting the workstation, the robot is guided by images and maps. After contacting the workstation, the robot immediately receives the contact signal of the electrode plate through the relay on the electrode plate. At this time, the high-precision ultrasonic function is used to detect the distance to the workstation and reverse the compression of the electrode plate. When the preset value is reached, the robot stops moving forward and calculates the actual current of the two motors at this time. The motor current and the compression of the electrode plate are then used to fit the data in the database to determine whether the robot has returned to the station normally. The database was created using a large amount of experimental data from the previous stage. The image recognition function performs image recognition through an image recognition model. The training method of the image recognition model includes: acquiring an initial training image, performing image signal processing augmentation on the initial training image, and obtaining a processed training image. The image recognition model is trained based on the processed training images; The ultrasonic recognition function includes acquiring ultrasonic detection signals during recognition.

2. The behavior-driven method for a robot returning to its workstation according to claim 1, characterized in that, The augmentation processing of the initial training image includes performing at least one of the following processing methods on the initial training image: RGB domain augmentation processing, HSV domain augmentation processing, and YUV domain augmentation processing.

3. The behavior-driven method for a robot returning to its workstation according to claim 2, characterized in that, The RGB domain augmentation process includes at least one of color information adjustment, gamma transformation, and random histogram equalization; wherein, the parameters in the color correction matrix and bias matrix used in color information adjustment are obtained by fine-tuning preset parameters based on random variables; the gamma coefficients used in gamma transformation are obtained by fine-tuning preset parameters based on random variables; and random histogram equalization refers to determining whether to perform histogram equalization based on random variables.

4. The behavior-driven method for a robot returning to its workstation according to claim 3, characterized in that, When the RGB domain augmentation process includes at least two of the following: color information adjustment, gamma transformation, and random histogram equalization, color information adjustment is performed before gamma transformation, and gamma transformation is performed before random histogram equalization.

5. The behavior-driven method for a robot returning to its workstation according to claim 2, characterized in that, The HSV domain augmentation process includes saturation adjustment and / or contrast adjustment; wherein, the saturation parameter used in the saturation adjustment is obtained by fine-tuning a preset parameter based on a random variable, and the contrast parameter used in the contrast adjustment is obtained by fine-tuning a preset parameter based on a random variable.

6. The behavior-driven method for a robot returning to its workstation according to claim 2, characterized in that, The YUV domain augmentation process includes YUV domain noise reduction and / or edge enhancement; wherein, the parameters of the low-pass filter used in YUV domain noise reduction are obtained by fine-tuning preset parameters based on random variables, and the parameters used in edge enhancement are obtained by fine-tuning preset parameters based on random variables.

7. The behavior-driven method for a robot returning to its workstation according to claim 1, characterized in that, The ultrasonic detection signal is an electrical signal obtained by converting the ultrasonic signal sensed by the ultrasonic sensor into an electrical signal. Before decomposing the ultrasonic detection signal into a specified number of signal components with equal bandwidth, the method further includes: filtering the ultrasonic detection signal to obtain a filtered ultrasonic detection signal; correspondingly, decomposing the ultrasonic detection signal into a specified number of signal components with equal bandwidth means: decomposing the filtered ultrasonic detection signal into a specified number of signal components with equal bandwidth.

Images