A capsule robot system with multi-target drug administration function

Abstract

The embodiment of this specification provides a capsule robot system with a multi-target drug delivery function, a capsule component, a moving part and a drug delivery part, an information collection unit, which records the movement process of the capsule robot and collects intra-cavity information in real time, a route planning unit, which performs reinforcement learning planning on the drug delivery sequence between multiple targets to obtain the movement route of the multiple targets, and a control unit, including: a parameter distribution module, an inclination adaptation module, and a speed adaptation module. The parameter distribution module locates according to the movement process and intra-cavity information, calculates the distribution value of the parameter to be adapted at the position, the inclination adaptation module adapts the inclination according to the distribution value, and the speed adaptation module adapts the speed according to the distribution value. The capsule robot moves along the movement route of the multiple targets by controlling the moving part, and the drug delivery part delivers the drug when it reaches the preset target position. The position is adapted through reinforcement learning and speed inclination, thereby improving the drug delivery efficiency and the safety of the cavity.

Description

Technical Field

[0001] The present application relates to the field of intelligent medical care, and in particular to a capsule robot system with multi-target drug delivery function. Background Art

[0002] As an actively controlled micro-robot, the capsule robot can safely and efficiently perform non-invasive diagnosis and treatment in cavities such as blood vessels, stomach, and intestines, avoiding the painful shortcomings of traditional gastroscopy diagnosis.

[0003] However, since the capsule robot moves inside a sensitive and fragile cavity, it is inevitable that it will cause damage. The current control of the capsule robot is mostly at a primary level, mostly involving only the principles of dynamic control. The control parameters of the movement process are mostly static values, which do not take into account the specific situation, resulting in damage to the cavity. Moreover, for multi-target drug administration, most of the time, the drugs are administered in a manually set order. In fact, different orders have different traumatic effects and efficiencies.

[0004] Therefore, it is necessary to provide a new method to improve the efficiency of drug application and the safety of the cavity. Summary of the Invention

[0005] The present invention provides a capsule robot system with a multi-target drug delivery function to improve drug delivery efficiency and cavity safety. The system specifically includes:

[0006] A capsule assembly having a traveling member and a drug dispensing member;

[0007] The information collection unit records the movement process of the capsule robot and collects the information inside the cavity in real time;

[0008] The route planning unit performs reinforcement learning planning on the dosing sequence among multiple targets to obtain the movement routes of multiple targets;

[0009] A control unit, comprising: a parameter distribution module, an inclination adaptation module, and a speed adaptation module. The parameter distribution module calculates a distribution value of the parameter to be adapted at the location of the capsule robot based on the motion process and the intracavity information positioning. The inclination adaptation module performs inclination adaptation based on the distribution value. The speed adaptation module performs speed adaptation based on the distribution value. The capsule robot moves along the motion route of the multiple targets and along the adapted speed and inclination by controlling the moving member.

[0010] The medicine dispensing member dispenses medicine when it moves to a preset target position.

[0011] Optionally, the step of performing reinforcement learning planning on the dosing sequence among multiple targets to obtain movement routes of the multiple targets includes:

[0012] Construct a three-dimensional model of the cavity and configure parameters according to the magnitude and direction of the fluid flow velocity in the cavity;

[0013] A medication route strategy search space is constructed based on the position of the capsule robot in the cavity, the next moving direction, and the target medication order, and a reward function is constructed. The indicators of the reward function include: the influence factor of the fluid flow rate on the speed of the capsule robot, the influence factor of the capsule robot's trauma to the cavity wall, and the total time consumed by the capsule robot. Multiple particle swarms are created to search for medication route strategies in the search space, and the reward function values are calculated respectively. The medication route strategy with the largest reward function value is selected, and a motion route is generated according to the position of the capsule robot in the cavity during the search process of the medication route strategy.

[0014] Optionally, the parameter distribution module is further used to:

[0015] The inverse of the closest vertical distance from the capsule robot to the tube wall is multiplied by the speed and then summed at each position to obtain the trauma impact factor.

[0016] Optionally, calculating the distribution value of the parameter to be adapted at the location of the capsule robot includes:

[0017] Using the preset distribution state prediction model, combined with the preset time consumption, the closest vertical distance from different positions to the pipe wall, the shape and texture of the cavity, and the fluidity parameters of the fluid, the corresponding distribution state is predicted. The distribution value corresponding to the position is determined based on the distribution state. The distribution state is the waveform of the parameter value to be adapted - the closest vertical distance.

[0018] Optionally, there is also a model training module for collecting the shape, texture, fluid flow parameters of the cavity, the speed, inclination angle of the robot, the closest vertical distance from each position to the tube wall and the degree of damage to the tube wall in historical drug application events. Training labels are set according to the degree of damage to the tube wall, the speed distribution and inclination angle distribution formed by the capsule robot at multiple positions. The total time taken by the capsule robot to move, the shape, texture and fluid flow parameters of the cavity are used as training sample inputs to train the pre-built model architecture and obtain a distribution state prediction model.

[0019] Optionally, setting training labels according to the degree of damage to the tube wall, the velocity distribution and the inclination angle distribution formed by the capsule robot at multiple positions includes:

[0020] The velocity distribution and inclination distribution of different historical pesticide application events were divided into black and white samples according to the degree of pipe wall trauma, and the velocity distribution segments and inclination distribution segments of the same historical pesticide application event were divided into black and white samples according to the degree of pipe wall trauma.

[0021] Optionally, the control unit further comprises:

[0022] The medicine application timing control unit is used to obtain the fluid pulsation timing information in the cavity, predict the robot displacement prediction pulsation value caused by the accumulation of the fluid pulsation state during the first moment in the future based on the fluid pulsation timing information, calculate the robot displacement control value during the first moment in the future based on the adapted speed value, and judge in real time whether the current moment has reached the medicine application time, specifically including: judging whether the vector of the robot displacement prediction pulsation value and the robot displacement control value is equal to the relative distance between the current position of the robot and the medicine application point; if so, it is determined that the medicine application time has arrived and the medicine application part is started, and the time difference between the first moment in the future and the current moment is equal to the time difference between starting the medicine application part and starting the medicine application part.

[0023] Optionally, the texture of the cavity includes: elasticity, toughness and thickness.

[0024] Optionally, the drug administration method is puncture.

[0025] Optionally, the fluid pulsation timing information includes a fluid velocity vector with time information, where positive and negative represent forward and backward.

[0026] The various technical solutions provided in the embodiments of this specification are provided through a capsule component, which has a moving part and a drug-dispensing part, an information collection unit, which records the movement process of the capsule robot and collects intra-cavity information in real time, a route planning unit, which performs reinforcement learning planning on the drug-dispensing sequence between multiple targets to obtain the movement route of multiple targets, and a control unit, including: a parameter distribution module, an inclination adaptation module, and a speed adaptation module. The parameter distribution module locates according to the movement process and intra-cavity information, calculates the distribution value of the parameter to be adapted at the position, the inclination adaptation module adapts the inclination according to the distribution value, and the speed adaptation module adapts the speed according to the distribution value. The moving part is controlled to make the capsule robot move along the movement route of multiple targets, and the drug-dispensing part dispenses the drug when it reaches the preset target position. The position is adapted through reinforcement learning and speed inclination, thereby improving the drug-dispensing efficiency and the safety of the cavity. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0028] Figure 1 This is a schematic diagram of the principle of a capsule robot system with multi-target drug delivery function provided in an embodiment of this specification. DETAILED DESCRIPTION

[0029] Exemplary embodiments of the present invention will now be described more fully with reference to the accompanying drawings. However, exemplary embodiments can be implemented in various forms, and it should not be understood that the present invention is limited to the embodiments set forth herein. On the contrary, providing these exemplary embodiments enables the present invention to be more comprehensive and complete, making it easier to fully convey the inventive concept to those skilled in the art. In the figures, the same reference numerals represent the same or similar elements, components or parts, and thus their repeated description will be omitted.

[0030] Under the premise of being consistent with the technical concept of the present invention, the features, structures, characteristics or other details described in a specific embodiment do not exclude that they can be combined in one or more other embodiments in a suitable manner.

[0031] In the description of specific embodiments, the features, structures, characteristics, or other details of the present invention are described to enable those skilled in the art to fully understand the embodiments. However, this does not preclude those skilled in the art from practicing the technical solutions of the present invention without one or more of the specific features, structures, characteristics, or other details.

[0032] The flowcharts shown in the accompanying drawings are for illustrative purposes only and do not necessarily include all contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may be decomposed, while others may be combined or partially combined. Therefore, the actual execution order may vary depending on the actual situation.

[0033] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically separate entities. That is, these functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0034] The term "and / or" or "and / or" includes all combinations of any one or more of the associated listed items.

[0035] Figure 1 The schematic diagram of the principle of a capsule robot system with multi-target drug delivery function provided in the embodiment of this specification may include:

[0036] A capsule assembly having a traveling member and a drug dispensing member;

[0037] The information collection unit 102 records the movement process of the capsule robot and collects the cavity information in real time;

[0038] The route planning unit 103 performs reinforcement learning planning on the dosing sequence among multiple targets to obtain the movement routes of the multiple targets;

[0039] The control unit 104 includes: a parameter distribution module 1041, an inclination adaptation module 1042, and a speed adaptation module 1043. The parameter distribution module calculates the distribution value of the parameter to be adapted at the location of the capsule robot based on the motion process and the intracavity information positioning. The inclination adaptation module performs inclination adaptation according to the distribution value. The speed adaptation module performs speed adaptation according to the distribution value. The capsule robot moves along the motion route of the multiple targets and at the adapted speed and inclination by controlling the moving member.

[0040] The medicine dispensing member dispenses medicine when it moves to a preset target position.

[0041] The system has a capsule component, a traveling part and a medicine-dispensing part, an information collection unit, which records the movement process of the capsule robot and collects intracavity information in real time, a route planning unit, which performs reinforcement learning planning on the medicine-dispensing sequence between multiple targets to obtain the movement route of multiple targets, and a control unit, which includes: a parameter distribution module, an inclination adaptation module, and a speed adaptation module. The parameter distribution module locates according to the movement process and intracavity information, calculates the distribution value of the parameter to be adapted at the position, the inclination adaptation module adapts the inclination according to the distribution value, and the speed adaptation module adapts the speed according to the distribution value. By controlling the traveling part, the capsule robot moves along the movement route of multiple targets, and the medicine-dispensing part dispenses medicine when it moves to the preset target position. The position is adapted through reinforcement learning and speed inclination, thereby improving the medicine-dispensing efficiency and the safety of the cavity.

[0042] This embodiment can be used in existing capsule robots of various mechanical structures, and can be upgraded to this system by modifying the program in the chip.

[0043] The positioning according to the movement process and the intracavity information may include:

[0044] Generate an acceleration time sequence and an acceleration reverse sequence based on the acceleration information collected by the gyroscope during the movement, input the acceleration time sequence and the acceleration reverse sequence into a pre-trained static state determination model, and output a static time sequence interval;

[0045] Calculate the velocity from the stationary interval and use it to calculate the displacement;

[0046] The three-dimensional coordinates of the robot are calculated by combining the intracavitary image and the displacement.

[0047] Among them, the acceleration time series of the sample can be collected and the static time series can be measured, the static time series interval can be marked on the acceleration time series, the acceleration time series can be reversed and matched with the acceleration time series, and the acceleration time series and reversed time series of the sample can be used to set training labels using the measured static time series. The acceleration time series and reversed time series can be used as a single training sample, and the static state judgment model can be trained using batch samples.

[0048] In this way, when determining the stationary state, the reverse order of acceleration can be taken into account, so the accuracy is higher.

[0049] Among them, the capsule component is the mechanical hardware of the capsule robot, the moving part is used to push the capsule robot to move and adjust the angle, and the drug application part is used to apply drugs to the target.

[0050] The information acquisition unit may include an image information acquisition unit, which acquires the position of the capsule robot by acquiring intracavity image information and identifying the image for positioning. It may also include an acceleration acquisition unit, which records the acceleration during movement and obtains the displacement through acceleration calculation to obtain the position of the capsule robot.

[0051] The route planning unit is used to plan the medication order and route among multiple targets, balancing the trauma impact and medication efficiency.

[0052] In the embodiments of this specification, the method of performing reinforcement learning planning on the administration sequence between multiple targets to obtain the movement routes of the multiple targets includes:

[0053] Construct a three-dimensional model of the cavity and configure parameters according to the magnitude and direction of the fluid flow velocity in the cavity;

[0054] A medication route strategy search space is constructed based on the position of the capsule robot in the cavity, the next moving direction, and the target medication order, and a reward function is constructed. The indicators of the reward function include: the influence factor of the fluid flow rate on the speed of the capsule robot, the influence factor of the capsule robot's trauma to the cavity wall, and the total time consumed by the capsule robot. Multiple particle swarms are created to search for medication route strategies in the search space, and the reward function values are calculated respectively. The medication route strategy with the largest reward function value is selected, and a motion route is generated according to the position of the capsule robot in the cavity during the search process of the medication route strategy.

[0055] The total time taken by the capsule robot refers to the total time taken from the start of movement to the completion of drug administration to multiple targets.

[0056] The factor affecting the velocity of the capsule robot by the fluid velocity may be a viscosity parameter of the fluid.

[0057] The control unit is used to calculate, judge and control the speed, inclination angle and drug administration timing of the capsule robot. It can also deploy various machine learning models to assist in calculations.

[0058] Through reinforcement learning, the optimal movement route can be learned, and the order of application, trauma impact and application efficiency can be balanced while taking into account various factors, thereby improving the scientificity, safety and application efficiency.

[0059] Considering that the greater the speed of the robot, the greater the risk of trauma, and the closer the robot is to the pipe wall when moving, the greater the risk of trauma, therefore, the trauma impact factor can be calculated by combining these two factors.

[0060] Therefore, in the embodiments of this specification, the parameter distribution module can also be used to:

[0061] The inverse of the closest vertical distance from the capsule robot to the tube wall is multiplied by the speed and then summed at each position to obtain the trauma impact factor.

[0062] Current technologies mostly use a constant speed to control the movement of capsule robots. However, to balance trauma impact and efficiency, different speeds can be configured at different locations. The closer to the tube wall, the slower the speed, thereby reducing trauma; the farther away from the tube wall, the faster the speed, thereby improving efficiency.

[0063] Due to the different characteristics of different cavity structures, the velocity attenuation amplitude can be predicted for each specific application situation. This attenuation trend is regarded as a distribution state and recorded as a waveform.

[0064] The inclination angle refers to the inclination angle between the working direction of the applicator and the normal direction of the pipe wall. The closer to the pipe wall, the smaller the inclination angle, which is more conducive to the application of the medicine. The farther away from the pipe wall, the larger the inclination angle, which is more conducive to the movement.

[0065] Therefore, in the embodiment of this specification, the calculation of the distribution value of the parameter to be adapted at the position where the capsule robot is located includes:

[0066] Using the preset distribution state prediction model, combined with the preset time consumption, the closest vertical distance from different positions to the pipe wall, the shape and texture of the cavity, and the fluidity parameters of the fluid, the corresponding distribution state is predicted. The distribution value corresponding to the position is determined based on the distribution state. The distribution state is the waveform of the parameter value to be adapted - the closest vertical distance.

[0067] The waveform of the parameter value to be adapted - the closest vertical distance is a function of x=X*(sigmod(d / D)-0.5), where the parameter to be adapted is x, d is the current closest vertical distance, D is the preset maximum closest vertical distance, and X is the preset maximum parameter to be adapted. X can be at least one of the inclination angle and the speed.

[0068] The cavity is a multi-dimensional closed figure and therefore often has multiple vertical distances. Since the closer the distance, the higher the risk of trauma, the closest vertical distance is selected as the evaluation indicator.

[0069] The fluidity parameters of a fluid can be obtained through measurement or prediction.

[0070] Different cavities can obtain waveforms that are adapted to them, so that the extent of the speed reduction in the process of the robot gradually approaching the pipe wall is adapted to the various properties of the cavity, thereby improving the overall travel efficiency and reducing the impact of trauma.

[0071] In the embodiments of this specification, the texture of the cavity includes: elasticity, toughness and thickness.

[0072] In an embodiment of the present specification, there is also a model training module for collecting the shape, texture, fluid flow parameters of the cavity, speed, inclination angle of the robot, the closest vertical distance from each position to the tube wall and the degree of damage to the tube wall in historical drug application events. Training labels are set according to the degree of damage to the tube wall, the speed distribution and inclination angle distribution formed by the capsule robot at multiple positions. The total time taken by the capsule robot to move, the shape, texture and fluid flow parameters of the cavity are used as training sample inputs to train the pre-built model architecture and obtain a distribution state prediction model.

[0073] The degree of trauma to the vessel wall can be quantified in advance, thereby achieving the above-mentioned training.

[0074] In the embodiment of this specification, the setting of training labels according to the degree of damage to the tube wall, the velocity distribution and the inclination distribution formed by the capsule robot at multiple positions includes:

[0075] The velocity distribution and inclination distribution of different historical pesticide application events were divided into black and white samples according to the degree of pipe wall trauma, and the velocity distribution segments and inclination distribution segments of the same historical pesticide application event were divided into black and white samples according to the degree of pipe wall trauma.

[0076] By dividing the segments, the upper limit of the speed allowed for the trauma results that meet the conditions at different locations can be accurately obtained, thereby improving the accuracy of the model prediction.

[0077] Considering that the fluid in the cavity often pulsates rhythmically, there is a short time difference between the activation of the drug dispensing component and the drug dispensing. This time difference will cause the displacement control value of the capsule robot to deviate from the actual displacement value. In order to reduce this deviation, it is necessary to combine this time difference to determine the accurate time of drug dispensing.

[0078] In the embodiment of this specification, the control unit further comprises:

[0079] The medicine application timing control unit is used to obtain the fluid pulsation timing information in the cavity, predict the robot displacement prediction pulsation value caused by the accumulation of the fluid pulsation state during the first moment in the future based on the fluid pulsation timing information, calculate the robot displacement control value during the first moment in the future based on the adapted speed value, and judge in real time whether the current moment has reached the medicine application time, specifically including: judging whether the vector of the robot displacement prediction pulsation value and the robot displacement control value is equal to the relative distance between the current position of the robot and the medicine application point; if so, it is determined that the medicine application time has arrived and the medicine application part is started, and the time difference between the first moment in the future and the current moment is equal to the time difference between starting the medicine application part and starting the medicine application part.

[0080] By judging whether the vector of the robot displacement prediction pulse value and the robot displacement control value is equal to the relative distance between the robot's current position and the application point, the constraint that the relative distance between the robot's current position and the application point must be equal to the preset distance before the application is done is eliminated, so that each relative distance may become the position of the robot when the application is made, thereby improving flexibility.

[0081] The velocity in the fluid pulsation timing information may be time-integrated to obtain a predicted pulsation value of the robot displacement caused by the accumulation, thereby characterizing the displacement caused by the fluctuation.

[0082] In the embodiments of this specification, the method of administering the medicine is a puncture method. Of course, it can also be other methods. In fact, the capsule robot can also perform microsurgery in addition to administering the medicine. There is no limitation here.

[0083] In the embodiment of this specification, the fluid pulsation timing information includes a fluid velocity vector with time information, where positive and negative represent forward and backward.

[0084] In a specific implementation, the fluid pulsation timing information can be expressed as a timing of speed changes with positive and negative signs.

[0085] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

[0086] The foregoing is merely an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A capsule robot system with multi-target drug delivery function, characterized in that: include: A capsule assembly having a traveling member and a drug dispensing member; The information collection unit records the movement process of the capsule robot and collects the information inside the cavity in real time; The route planning unit performs reinforcement learning planning on the dosing sequence among multiple targets to obtain the movement routes of multiple targets; The method of performing reinforcement learning planning on the administration sequence among multiple targets to obtain the movement routes of the multiple targets includes: Construct a three-dimensional model of the cavity and configure parameters according to the magnitude and direction of the fluid flow velocity in the cavity; A medication route strategy search space is constructed based on the position of the capsule robot in the cavity, the next movement direction, and the target medication sequence, and a reward function is constructed. The indicators of the reward function include: the influence factor of the fluid flow rate on the capsule robot's speed, the influence factor of the capsule robot's trauma to the cavity wall, and the total time consumed by the capsule robot. Multiple particle swarms are created to search for medication route strategies in the search space, and the reward function values are calculated for each. The medication route strategy with the largest reward function value is selected, and a motion route is generated based on the position of the capsule robot in the cavity during the search process of the medication route strategy. A control unit, comprising: a parameter distribution module, an inclination adaptation module, and a speed adaptation module. The parameter distribution module calculates a distribution value of the parameter to be adapted at the location of the capsule robot based on the motion process and the intracavity information positioning. The inclination adaptation module performs inclination adaptation based on the distribution value. The speed adaptation module performs speed adaptation based on the distribution value. The capsule robot moves along the motion route of the multiple targets and along the adapted speed and inclination by controlling the moving member. The control unit further comprises: a medication application timing control unit, configured to obtain fluid pulsation timing information in the cavity, predict a predicted robot displacement pulsation value resulting from the accumulation of fluid pulsation states during a first future moment based on the fluid pulsation timing information, calculate a robot displacement control value during a first future moment based on an adapted speed value, and determine in real time whether the medication application timing has been reached at the current moment, specifically comprising: determining whether the vector sum of the robot displacement predicted pulsation value and the robot displacement control value is equal to the relative distance between the robot's current position and a medication application point; if so, determining that the medication application timing has been reached and activating the medication application component, wherein the time difference between the first future moment and the current moment is equal to the time difference between activating the medication application component and starting the medication application component; The parameter distribution module is further used to: The inverse of the closest vertical distance between the capsule robot and the tube wall is multiplied by the speed and then summed at each position to obtain the trauma impact factor. The calculating the distribution value of the parameter to be adapted at the location of the capsule robot includes: Using a preset distribution state prediction model, combined with preset time consumption, the closest vertical distance from different positions to the pipe wall, the shape and texture of the cavity, and the fluidity parameters of the fluid, a distribution state corresponding to the position is predicted. Based on the distribution state, the distribution value corresponding to the position is determined. The distribution state is a waveform of the parameter value to be adapted - the closest vertical distance; The medicine dispensing member dispenses medicine when it moves to a preset target position; It also has a model training module for collecting the shape, texture, fluid flow parameters of the cavity, the speed, inclination angle of the robot, the closest vertical distance from each position to the tube wall and the degree of damage to the tube wall in historical drug application events. Training labels are set according to the degree of damage to the tube wall, the speed distribution and inclination angle distribution formed by the capsule robot at multiple positions. The total time taken by the capsule robot to travel, the shape, texture and fluid flow parameters of the cavity are used as training sample inputs to train the pre-built model architecture and obtain a distribution state prediction model.

2. The system according to claim 1, wherein: The step of setting training labels according to the degree of damage to the tube wall, the velocity distribution formed by the capsule robot at multiple positions, and the inclination distribution includes: The velocity distribution and inclination distribution of different historical pesticide application events were divided into black and white samples according to the degree of pipe wall trauma, and the velocity distribution segments and inclination distribution segments of the same historical pesticide application event were divided into black and white samples according to the degree of pipe wall trauma.

3. The system according to claim 2, characterized in that The texture of the cavity includes elasticity, toughness and thickness.

4. The system according to claim 1, wherein: The method of administering the medicine is piercing.

5. The system according to claim 1, wherein: The fluid pulsation timing information includes a fluid velocity vector with time information, where positive and negative represent forward and backward.

Images