Home service robot for home-based care for aged

Abstract

The invention relates to a home-based care home service robot which is characterized by sequentially comprising a cleaning base, a main body and a head with a touch screen from bottom to top. The main body comprises a plurality of dinner plates capable of being stored, the main body is internally provided with a partition bin independently provided with a main board, and when the dinner plates are stored, heating and / or heat preservation can be carried out in the main body, and a path for reaching the service object is found in the path according to the position of the current robot so that the robot can walk to the side of the service object. By means of path calculation and long and short memory model training, the robot can intelligently and regularly carry out operation and article delivery on household articles and cared objects and carry out voice communication, and efficient assistance is achieved for life of the old.

Description

Technical Field

[0001] The present invention relates to a household service robot, in particular to a household service robot for elderly care at home, and belongs to the field of intelligent service robots. Background Art

[0002] Existing service robots are dual infrastructures for data storage and remote control, or are equipped with functions that can deliver items (such as delivering food and water, etc.), and are robots that provide services on demand. However, the reality is that when the elderly cannot operate the robot, or even lose the ability to express and communicate, they still need to be unattended, especially before the development of assistance in the use of household items (such as cooking, cleaning, etc.) and active movement control of the elderly's limbs, such as supporting, lifting, moving, temporarily helping to maintain a posture, or even feeding, going to the toilet, bathing, etc., how can robots intelligently understand the living habits of different elderly people, pay attention to the physical and mental dynamics of the elderly, and dynamically make auxiliary service strategies that are still mainly based on people. Existing technologies do not take this into account. Summary of the invention

[0003] In view of the above-mentioned problems in the prior art, the present invention considers the following aspects: first, basic functions of cleaning and delivering items are introduced into the robot; second, intelligent algorithms are used to learn the living habits of the elderly, so as to adopt corresponding service strategies; third, remote communication with personnel is carried out to collect corresponding manual services, saving manual effort.

[0004] Based on the above considerations, the present invention provides a home-based elderly care service robot, comprising, from bottom to top, a cleaning base, a main body, and a head with a touch screen, wherein: The cleaning base includes a shell, a cleaning structure arranged inside the shell, a walking mechanism, a dust collecting bin, a water tank, a dirt suction box, a plurality of camera devices arranged on the side of the cleaning base, and a first intelligent chip, which is used to perform three-dimensional modeling and path calculation according to the images taken in real time by the camera devices. The main body includes a plurality of storable dish plates, and a compartment with a separate main board is provided inside the main body. When the dish plates are stored, they can be heated and / or kept warm inside the main body, and a path to the service object can be found in the path according to the current position of the robot and the robot can walk to the service object. A second intelligent chip is arranged on the main board, which can perform cleaning work at a fixed time, remotely control the household appliances required by the service object at a fixed time, and communicate remotely with personnel through a pre-trained model to remind manual service time and the current physiological state of the service object; the personnel can modify the time of cleaning work and remote control operation of household appliances through the touch screen, and the main board regularly retrains the model according to the modified historical data, and when the timing result obtained based on the newly trained model is inconsistent with the last modification record, the personnel is remotely informed to manually determine whether to select the timing result obtained by the newly trained model or the last modification record, so as to remotely execute the remote control operation and / or cleaning work of household appliances.

[0005] Optionally, the main board has a voice dialogue module, which can recognize the voice of the service object and the voice of the personnel based on the collected voice, so that when the service object and / or the personnel have a voice dialogue, it has priority over the pre-trained model and the remote execution to perform the cleaning work and / or remote control operation.

[0006] Optionally, the human voice conversation includes on-site conversation and remote conversation.

[0007] Optionally, the cleaning base includes a universal wheel controlled by a main board, which performs dry sweeping first and then wet sweeping, and the cleaning structure includes a roller brush that performs both dry sweeping and wet sweeping. The water tank is connected to a water injection pipe through an electronic valve controlled by the main board, and water is introduced into the surface of the roller brush for wetting. The dust collecting bin is located above the roller brush, and a shovel head extends from one end of the dust collecting bin to shovel debris off the roller brush and continuously accumulate it in the dust collecting bin. A water sump is formed between the suction box and the roller brush, and the sewage in the water sump is sucked out through a water pump on one side of the water tank and also arranged in the outer shell and controlled by the main board through a water pump pipe. The water pump pipe starts from the water sump and passes through the suction box to connect the water pump pipe, and the water pump pipe has an opening in a section passing through the interior of the suction box to allow the sucked water to overflow and fall into the suction box, and the water outlet of the water pump returns water to the suction box.

[0008] Preferably, a filter is installed at the joint where the water suction pipe is connected to the water suction pipe and / or at the return water section that penetrates into the sewage suction box.

[0009] Optionally, a distance sensor is installed on the camera device to define an initial position, so that the universal wheel controls the cleaning base to perform a first rotation, collects images taken by the camera device before and after the first rotation, and distance data collected during the first rotation, and continues to complete the second rotation, completes one circle of rotation, and also continues to collect distance data collected during the second rotation, and the first intelligent chip performs three-dimensional modeling based on the image and all distance data.

[0010] Preferably, the plurality of camera devices are equally distributed on a circle, and the first rotation angle is half of the acute center angle between the positions of two adjacent camera devices and the center of the circle.

[0011] Optionally, the step of performing three-dimensional modeling according to the image and all distance data comprises: S1 establishes a three-dimensional rectangular coordinate system in the modeling software, locates all distance data one by one based on the center position of the base circle, and connects each positioning point in sequence; S2, in a three-dimensional rectangular coordinate system, identifies whether the image belongs to furniture based on the images before and after the first rotation, locates the furniture according to the center position of the photographed circle, and establishes a three-dimensional body of the furniture according to a preset height estimation value of the furniture; S3 determines the projection contour of the furniture 3D body on the ground, and establishes an initial path in the non-furniture projection contour area, and walks along the initial path. After every preset time period, stop and continue with steps S1-S2, and continue to establish more paths until the scanning is completed to obtain all non-furniture projection contour areas; S4 finds the nearest path to the vicinity of the current location of the service object among all established paths, and defines the path as the fastest arrival path based on the current location.

[0012] The establishment of the initial path and other paths is determined by the user within the non-furniture projection outline area displayed on the touch screen or smart phone, and after determination, the main board controls the universal wheels according to the path to make the robot walk.

[0013] It is easy to understand that, due to the establishment of the three-dimensional rectangular coordinate system and the determination of the current center position of the base, it is easy to determine the walking parameters based on the equation of the path.

[0014] Preferably, after the path is established, if the mainboard finds through calculation that walking along the established path will cause collision based on the current furniture and non-furniture projection contour areas, the user will be asked to re-determine a new path until the calculation result shows that no collision will occur.

[0015] It can be understood that according to the maximum distance in the shoulder width direction of the robot, the distance between the symmetrical points on both sides of the vertical path is calculated to see whether it is greater than the distance in the shoulder width direction, so as to determine whether there is a collision. It is generally best to leave an extra distance of 2-5cm just in case.

[0016] Optionally, the user can set a departure point for the robot, and when all tasks are completed, the robot will return to the departure point and wait for further orders.

[0017] Optionally, the method for pre-training the model includes the following steps: Q1 obtains the touch screen on-site user settings, voice history records, and remote modification timestamps, and divides the same type of control commands every day into training sets and verification sets; Q2 builds a long-short memory model, inputs the control instruction training set into different model network units in chronological order, calculates the loss function, continuously optimizes the network parameters, and uses the validation set for verification. When the loss function value is the smallest, stop training and obtain the pre-trained model.

[0018] Optionally, a method of walking to the service object and finding a path to the service object based on the current position of the robot is to connect the current position with the nearest starting point of the path in step S4 in a straight line. When the main board calculates that there will be a collision in the straight line, the user will adjust it through the touch screen or the remote smartphone until it is calculated that there will be no collision. Beneficial Effects

[0019] By utilizing path calculation and long-short memory model training, the robot can intelligently and regularly operate and deliver household items and cared-for objects, as well as perform voice communication, thus providing efficient assistance to the daily lives of the elderly. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A schematic diagram of the structure of a home-based elderly care service robot according to the present invention, including a bottom view of the base, a side view, and a side view of the main body and a part of the base shell. Figure 2 Schematic diagram of the position change of the camera device before and after the first rotation and the range focus of the front and rear fields of view, Figure 3 3D floor modeling and the nearest path diagram, Figure 4 A simplified diagram of long short-term memory model training. DETAILED DESCRIPTION

[0021] Figure 1 The present invention provides a home-based elderly care service robot, which includes, from bottom to top, a cleaning base, a main body, and a head with a touch screen, wherein: The cleaning base includes a T-shaped shell, a cleaning structure arranged inside the T-shaped shell, universal wheels, a dust collection bin, a water tank, a dirt suction box, a plurality of camera devices arranged on the side of the cleaning base, and a first intelligent chip, which is used to perform three-dimensional modeling and path calculation according to the image taken in real time by the camera device. The main body includes 4 storable dish trays, and the main body has a compartment with a separate main board. When the dish trays are stored, the main body can be heated and kept warm inside, and a path to the service object can be found in the path according to the current position of the robot and the robot can walk to the service object. A second intelligent chip (not shown in the figure) is arranged on the main board, which can perform cleaning work at a fixed time, remotely control the household appliances required by the service object at a fixed time, and communicate remotely with personnel through a pre-trained model to remind manual service time and the current physiological state of the service object; the personnel can modify the time of cleaning work and remote control operation of household appliances through the touch screen, and the main board regularly retrains the model according to the modified historical data, and when the timing result obtained based on the newly trained model is inconsistent with the last modification record, the personnel is remotely informed to manually determine whether to select the timing result obtained by the newly trained model or the last modification record, so as to remotely execute the remote control operation and cleaning work of household appliances.

[0022] Specifically, the main board also has a voice dialogue module, which can recognize the voice of the service object and the voice of the personnel based on the collected voice, so that in the on-site or remote mode, when the service object and the personnel have a voice dialogue, the priority is higher than the model and the remote execution to perform the cleaning work and remote control operation.

[0023] The cleaning structure includes Figure 1 The middle side view shows the universal wheel controlled by the main board, which first performs dry sweeping and then wet sweeping. The cleaning structure includes a roller brush that performs both dry sweeping and wet sweeping. The water tank is connected to the water injection pipe through an electronic valve controlled by the main board, and water is introduced into the surface of the roller brush for wetting. The dust collecting bin is located above the roller brush, and a shovel head extends from one end of the dust collecting bin to shovel off the debris on the roller brush and continuously accumulate in the dust collecting bin. A water collection tank is formed between the suction box and the roller brush. The sewage in the water collection tank is sucked out through a water pump controlled by the main board on one side of the water tank. The water pump starts from the water collection tank and passes through the suction box to connect the water pump, and the water pump has an opening in a section passing through the interior of the suction box to allow the sucked water to overflow and fall into the suction box. The water outlet of the water pump returns water to the suction box. A filter (not shown in the figure) is installed at the joint where the water pump is connected to the water pump and / or the return water section that penetrates into the suction box.

[0024] like Figure 1 , in the front and back indication states, the control lines of the water pump and the electronic valve are designed to pass through the outer shell and connect to the main board inside the main body. The dish is revolving or drawer-type and can be put in or pulled out from inside and outside the main body. A camera is also provided on the head to identify the front face image of the person being cared for so that AI can be used to turn it into a dynamic picture. This can comfort some patients with severe dementia who can mistake the robot for their relatives, thereby increasing the quality of service. A rotating disk is provided between the main body and the T-shaped shell so that the camera can easily find and align the collected face.

[0025] The heating and keeping warm of the food on the plate can be set via the touch screen.

[0026] like Figure 2 As shown, four equally divided camera devices are equipped with distance sensors (not shown in the figure), and the initial position A is defined, so that the universal wheel controls the cleaning base to perform the first rotation, that is, a 45° clockwise rotation. The four camera devices collect the images taken by the camera devices before and after the first rotation, as well as the distance data collected during the first rotation, and continue to complete the second rotation, complete a circle of rotation, and continue to collect the distance data collected during the second rotation. The first smart chip performs three-dimensional modeling based on the images and all the distance data. Therefore, the field of view of the adjacent camera devices and the focus of the camera device after the first rotation are P and P', so as to capture the images of the four sides of the circle O as much as possible.

[0027] Among them, Figure 3 As shown, the steps of performing three-dimensional modeling based on the image and all distance data include: S1 establishes a three-dimensional rectangular coordinate system (with its XY plane built on the floor) in the modeling software, locates all distance data one by one based on the center position of the base circle, and connects each positioning point in sequence; in the figure, starting from position a, the distance sensor scans the walls of the living room and kitchen, as well as the feet of the sofa and TV cabinet. The unscanned areas are still represented by thin lines.

[0028] S2 is a three-dimensional rectangular coordinate system, in which the image before and after the first rotation ( Figure 2 ), identify whether it is furniture, locate the furniture according to the center of the circle shot, and build the furniture 3D according to the estimated value of the preset height of the furniture (usually no more than 90cm, even if the height of the furniture exceeds 90cm, it will not affect the recognition); S3 has three bodies projected on the ground to obtain the furniture projection outline, and establish an initial path in the non-furniture projection outline area, and walk along the initial path, stop at position b in the figure every preset time period (usually 2-4s), at this time, continue with steps S1-S2, and continue to establish more paths until the scanning is completed, and all non-furniture projection outline areas are obtained; S4 finds the shortest path to the current location of the service object among all established paths, and defines the path as the fastest arrival path based on the current location. The path indicated in the figure is the shortest path, which leads to the bedside.

[0029] After the path is established, if the main board finds through calculation that a collision will occur if the robot walks along the established path based on the current furniture and non-furniture projection contour areas, the user will be asked to re-determine a new path until the calculation result shows that there will be no collision. Moreover, the establishment of the initial path and other paths is determined by the user within the non-furniture projection contour area displayed on the touch screen or smart phone, and after the determination, the main board controls the universal wheels according to the path to make the robot walk.

[0030] like Figure 4 As shown, the method of pre-training the model includes the following steps: Q1 obtains the touch screen on-site user settings, voice history records, and remote modification timestamps, and divides the same type of control commands every day into training sets and verification sets; Q2 builds a long-short memory model, inputs the control instruction training set into different model network units in chronological order, calculates the loss function, continuously optimizes the network parameters, and uses the validation set for verification. When the loss function value is the minimum, stop training and obtain the pre-trained model. Figure 4 Simply replace the training set with the validation set.

[0031] The method of finding a path to reach the service object by walking according to the current position of the robot is to connect the straight line ca with the nearest starting point of the path in step S4 based on the current position to the kitchen entrance c. When the main board calculates that there will be a collision in the straight line ca, the user adjusts it through the touch screen or the remote smart phone until it is calculated that there will be no collision.

Claims

1. A home-based elderly care service robot, characterized in that: The invention comprises, from bottom to top, a cleaning base, a main body, and a head with a touch screen, wherein: The cleaning base includes a shell, a cleaning structure arranged inside the shell, a walking mechanism, a dust collecting bin, a water tank, a dirt suction box, a plurality of camera devices arranged on the side of the cleaning base, and a first intelligent chip, which is used to perform three-dimensional modeling and path calculation according to the images taken in real time by the camera devices. The main body includes a plurality of storable dish plates, and a compartment with a separate main board is provided inside the main body. When the dish plates are stored, they can be heated and / or kept warm inside the main body, and a path to the service object can be found in the path according to the current position of the robot and the robot can walk to the service object. A second intelligent chip is arranged on the main board, which can perform cleaning work at a fixed time, remotely control the household appliances required by the service object at a fixed time, and communicate remotely with personnel through a pre-trained model to remind manual service time and the current physiological state of the service object; the personnel can modify the time of cleaning work and remote control operation of household appliances through the touch screen, and the main board regularly retrains the model according to the modified historical data, and when the timing result obtained based on the newly trained model is inconsistent with the last modification record, the personnel is remotely informed to manually determine whether to select the timing result obtained by the newly trained model or the last modification record, so as to remotely execute the remote control operation and / or cleaning work of household appliances.

2. The robot according to claim 1, characterized in that: The main board has a voice dialogue module, which can recognize the voice of the service object and the voice of the personnel according to the collected voice, so that when the service object and / or the personnel have a voice dialogue, the priority is higher than the pre-trained model and the remote execution to perform the cleaning work and / or remote control operation; Human voice conversations include on-site conversations and remote conversations.

3. The robot according to claim 1, characterized in that: The cleaning base includes a universal wheel controlled by a main board, which first performs dry sweeping and then wet sweeping. The cleaning structure includes a roller brush that performs both dry sweeping and wet sweeping. The water tank is connected to a water injection pipe through an electronic valve controlled by the main board, and water is introduced into the surface of the roller brush for wetting. The dust collecting bin is located above the roller brush, and a shovel head extends from one end of the dust collecting bin to shovel debris on the roller brush and continuously accumulate it in the dust collecting bin. A water collecting tank is formed between the suction box and the roller brush. The sewage in the water collecting tank is sucked out through a water pump controlled by the main board and located on one side of the water tank in the shell. The water pump starts from the water collecting tank and passes through the suction box to connect the water pump. The water pump has an opening in a section passing through the interior of the suction box to allow the sucked water to overflow and fall into the suction box. The water outlet of the water pump returns water to the suction box. A filter is installed at the joint where the water suction pipe is connected to the water suction pipe and / or the return water section that penetrates into the dirt suction box.

4. The robot according to any one of claims 1 to 3, characterized in that: A distance sensor is installed on the camera device to define an initial position, so that the universal wheel controls the cleaning base to perform a first rotation, collects images taken by the camera device before and after the first rotation, and distance data collected during the first rotation, and continues to complete the second rotation, completes one circle of rotation, and also continues to collect distance data collected during the second rotation. The first intelligent chip performs three-dimensional modeling based on the image and all distance data.

5. The robot according to claim 4, characterized in that: The plurality of camera devices are equally distributed on a circle, and the first rotation angle is half of the acute center angle between the positions of two adjacent camera devices and the center of the circle.

6. The robot according to claim 5, characterized in that: The steps of performing three-dimensional modeling based on the image and all distance data include: S1 establishes a three-dimensional rectangular coordinate system in the modeling software, locates all distance data one by one based on the center position of the base circle, and connects each positioning point in sequence; S2, in a three-dimensional rectangular coordinate system, identifies whether the image belongs to furniture based on the images before and after the first rotation, locates the furniture according to the center position of the photographed circle, and establishes a three-dimensional body of the furniture according to a preset height estimation value of the furniture; S3 determines the projection contour of the furniture 3D body on the ground, and establishes an initial path in the non-furniture projection contour area, and walks along the initial path. After every preset time period, stop and continue with steps S1-S2, and continue to establish more paths until the scanning is completed to obtain all non-furniture projection contour areas; S4 finds the nearest path to the vicinity of the current location of the service object among all established paths, and defines the path as the fastest arrival path based on the current location.

7. The robot according to claim 6, characterized in that: The establishment of the initial path and other paths is determined by the user in the non-furniture projection outline area displayed on the touch screen or smart phone, and after the determination, the main board controls the universal wheels according to the path to make the robot walk; After the path is established, the mainboard calculates and finds that walking along the established path will cause collisions based on the current furniture and non-furniture projection contour areas. It will then ask the user to re-determine a new path until the calculation result shows that there will be no collisions. Users can set the robot's departure point, and when all tasks are completed, it will return to the departure point and wait for orders.

8. The robot according to claim 7, characterized in that: The method of pre-training the model includes the following steps: Q1 obtains the touch screen on-site user settings, voice history records, and remote modification timestamps, and divides the same type of control commands every day into training sets and verification sets; Q2 builds a long-short memory model, inputs the control instruction training set into different model network units in chronological order, calculates the loss function, continuously optimizes the network parameters, and uses the validation set for verification. When the loss function value is the smallest, stop training and obtain the pre-trained model.

9. The robot according to claim 8, characterized in that: The method of finding a path to reach the service object according to the current position of the robot and walking to the service object is to connect the current position with the nearest path starting point in step S4 in a straight line. When the main board calculates that there will be a collision in the straight line, the user will adjust it through the touch screen or the remote smartphone until it is calculated that there will be no collision.

10. The method according to claim 8, characterized in that The head is also provided with a camera for identifying the frontal facial image of the person being cared for so as to convert it into a dynamic picture using AI, wherein a rotating disk is provided between the main body and the T-shaped shell so that the camera can easily find and align to capture the face; the heating and heat preservation are set through the touch screen.

Images