A multi-modal perception and memory alloy driven intelligent bathing robot

Abstract

The application relates to the technical field of intelligent nursing, and specifically discloses an intelligent bathing robot based on multi-modal sensing and memory alloy driving, which comprises a multi-degree-of-freedom cooperative cleaning system, an intelligent control core and a multifunctional integrated module. The multi-degree-of-freedom cooperative cleaning system comprises a flexible scrubbing arm, a spraying arm and a composite joint. The flexible scrubbing arm and the spraying arm are connected through the composite joint. The composite joint is internally provided with a distributed temperature sensor and a strain feedback unit, which are used for dynamically compensating motion errors. The intelligent control core comprises a multi-modal sensing module, a motion planning module and a dynamic optimization module. The dynamic optimization module is used for real-time adjustment of scrubbing intensity, spraying intensity and massage parameters. The application solves the problems of poor flexibility, poor environmental adaptability and high energy consumption of traditional bathing robots, realizes a technical breakthrough in safety, intelligence and user experience, and provides an efficient and comfortable integrated solution for the field of old-age nursing and rehabilitation aids.

Description

Technical Field

[0001] This invention relates to the field of intelligent nursing technology, specifically to an intelligent bathing robot based on multimodal perception and memory alloy drive. Background Technology

[0002] Bathing, as a crucial part of daily hygiene and care, is of paramount importance to the elderly. It cleanses the body, promotes metabolism, and aids in disease prevention. However, as people enter old age, both their physiological functions and psychological well-being face a degree of decline. Focusing on disabled elderly people in my country, the need for bathing assistance ranks first among the six daily activity needs. However, the slippery conditions of bathroom floors, coupled with the aging organs and difficulty in changing positions among disabled and semi-disabled elderly individuals, make them prone to slips, falls, and even suffocation during bathing, potentially leading to serious complications such as head injuries and fractures.

[0003] While current intelligent bathing robots on the market have solved some problems to a certain extent, they still have many shortcomings. For example, the multi-shower arm intelligent bathing robot disclosed in invention patent CN114504257A uses servo motors to control the movement of the scrubbing arms. Although it can achieve multi-angle adjustment, the large inertia and delayed response of the motor-driven joints can lead to fluctuations in scrubbing contact force, easily causing the risk of skin pressure sores. The intelligent massage bathing robot disclosed in invention patent CN221730388U, although its screw-driven wiping module has basic cleaning functions, its linear motion trajectory is fixed and it is difficult to adapt to the complex curves of the human body (such as the shoulders, neck, waist, and back). In addition, existing bathing robots rely on gearboxes or hydraulic systems, which have complex mechanical structures, high operating noise, and continuous power consumption, making it difficult to meet the needs of long-term care scenarios.

[0004] In summary, to address the problems of insufficient compliance, poor environmental adaptability, and high energy consumption of existing bathing robots, this invention proposes an intelligent bathing robot based on multimodal perception and memory alloy actuation. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent bathing robot based on multimodal perception and memory alloy drive, which aims to improve the problems of insufficient compliance, poor environmental adaptability and high energy consumption of existing bathing robots.

[0006] This invention is implemented as follows:

[0007] A smart bathing robot based on multimodal sensing and shape memory alloy drive includes:

[0008] A multi-degree-of-freedom collaborative cleaning system, specifically including:

[0009] A flexible scrubbing arm with an integrated MEMS pressure sensor at its end for real-time monitoring of scrubbing contact force;

[0010] The spray arm, controlled by a shape memory alloy wire drive unit, can achieve adaptive adjustment of height and angle through current pulse modulation. The end of the spray arm is equipped with a rotatable nozzle.

[0011] The flexible scrubbing arm and the spray arm are connected by a composite joint, which has a built-in distributed temperature sensor and a strain feedback unit for dynamically compensating for motion errors.

[0012] The intelligent control center specifically includes:

[0013] The multimodal perception module integrates a binocular vision unit, a 16-line LiDAR, and a voice interaction module.

[0014] The motion planning module is used to generate scrubbing trajectories that adapt to the curved surfaces of the human body;

[0015] The dynamic optimization module is used to adjust the scrubbing intensity, spray strength, and massage parameters in real time.

[0016] The multi-functional integrated module specifically includes:

[0017] Hot air drying unit for human body drying;

[0018] Embedded massage mechanism for human body massage;

[0019] The data fusion unit integrates data from multiple sensors using a Kalman filter algorithm to achieve coordinated control of bathing, massage, and nursing actions.

[0020] Furthermore, the drive structure of the flexible scrubbing arm includes:

[0021] An array of shape memory alloy wires consisting of multiple sets of nickel-titanium alloy wires, wherein the alloy wires are spirally wound on the surface of a flexible skeleton and each is connected to an independent temperature control circuit.

[0022] A conical helical return spring, which together with the memory alloy wire array forms an antagonistic drive unit.

[0023] Furthermore, the coordinated control method for the spray arm includes:

[0024] A three-dimensional human body model is constructed based on the sensing data from the multimodal sensing module, and the coverage area of ​​each spray arm is calculated.

[0025] A game theory optimization model is used to allocate spraying tasks and adjust the posture of the spraying arms;

[0026] And the spray flow rate is controlled by pulse width modulation.

[0027] Furthermore, the motion planning module employs an improved RRT algorithm and incorporates human body curvature constraints to generate a scrubbing trajectory adapted to the human body's curved surface; specific steps include:

[0028] Introduce human body surface curvature constraints into the standard RRT algorithm, and generate a smooth scrubbing path by fitting B-spline curves.

[0029] A step size strategy that is adaptively adjusted based on the radius of curvature is adopted;

[0030] In the path optimization stage, a contact force weighting factor is added to achieve real-time coupling of trajectory planning and pressure feedback.

[0031] Furthermore, the dynamic optimization module adjusts the scrubbing force, spray intensity, and massage parameters in real time through a fuzzy PID controller. The fuzzy PID controller takes contact force deviation, posture deviation, and water flow pressure as input variables and outputs drive current, spray flow rate, and vibration frequency.

[0032] The fuzzy PID controller uses contact force deviation, deviation change rate and water flow pressure as input variables, and shape memory alloy drive current, spray valve opening degree and vibration frequency as output variables, and includes a preset fuzzy rule base.

[0033] Furthermore, the hot air drying unit includes an infrared thermometer for real-time monitoring of skin surface temperature, a fan and heating structure for generating hot air, and a temperature controller. The infrared thermometer, fan, and heating structure are all electrically connected to the temperature controller. The hot air drying unit adopts a feedforward-feedback composite control strategy and is equipped with multiple drying modes, including a fast drying mode and a comfortable drying mode. The set temperature of the fast drying mode is higher than that of the comfortable drying mode, and the airflow speed of the fast drying mode is higher than that of the comfortable drying mode.

[0034] Furthermore, the embedded massage mechanism can execute a low-frequency deep pressure mode and a high-frequency kneading mode, and automatically matches the massage mode according to the contact force distribution fed back by the MEMS pressure sensor: when the pressure is greater than the set pressure threshold, it switches to the low-frequency deep pressure mode, and when the pressure is less than or equal to the set pressure threshold, it switches to the high-frequency kneading mode; the frequency range of the low-frequency deep pressure mode is 10-30Hz, and the frequency range of the high-frequency kneading mode is 50-100Hz; the massage path of the embedded massage mechanism is generated based on an improved RRT algorithm and is executed alternately with the wiping action, and the time interval can be set.

[0035] Furthermore, an inertial measurement unit is provided within the composite joint or at the end of the flexible scrubbing arm, and the workflow of the data fusion unit includes:

[0036] Synchronize binocular vision, lidar, and inertial measurement unit data using timestamps;

[0037] An extended Kalman filter algorithm is used to fuse information from multiple sources and output a 6-DOF pose estimate.

[0038] When a sudden human movement is detected, an emergency braking protocol is triggered. The sudden movement refers to the absolute value of acceleration collected in real time by the inertial measurement unit exceeding the set acceleration threshold.

[0039] Furthermore, the voice interaction module integrates a voice command recognition unit and a voiceprint recognition unit. The voice command recognition unit has a voice command set, which includes multiple control commands. The voiceprint recognition unit can be bound to multiple users, and each user has a corresponding personalized bathing plan. The personalized bathing plan includes, but is not limited to, scrubbing contact force, spray water temperature and flow rate, drying mode, and massage mode.

[0040] Furthermore, it also includes a safety protection module, which includes a dielectric elastomer insulating layer covering the surface of the flexible scrubbing arm, a redundant emergency stop circuit for cutting off the power and activating mechanical self-locking under abnormal conditions, and a wireless communication module for status uploading and remote intervention; the breakdown voltage of the dielectric elastomer insulating layer is ≥5kV; the abnormal conditions include when the contact force exceeds the set warning pressure threshold and when the temperature exceeds the set abnormal temperature threshold.

[0041] Compared with the prior art, the beneficial effects of the present invention are: the present invention solves the problems of insufficient compliance, poor environmental adaptability, high energy consumption and single function of traditional bathing robots by biomimetic drive design, multi-source perception fusion and closed-loop control optimization. It achieves technological breakthroughs in safety, intelligence and user experience, and provides an efficient and comfortable integrated solution for the field of elderly care and rehabilitation aids. Attached Figure Description

[0042] Figure 1 This is a block diagram of the module composition and a schematic diagram of the electrical connection structure of the present invention;

[0043] Figure 2 This is a block diagram showing the modular composition of the composite joint of the present invention;

[0044] Figure 3 This is a block diagram showing the modular composition of the intelligent control center of the present invention;

[0045] Figure 4 This is a block diagram showing the module composition of the safety protection module of the present invention;

[0046] Figure 5 This is a schematic diagram of a drive structure for one section of the flexible scrubbing arm of the present invention.

[0047] In the diagram: 100, Flexible scrubbing arm; 101, Flexible skeleton; 102, Shape memory alloy wire array; 103, Conical spiral return spring; 200, Spray arm; 300, Composite joint; 301, Distributed temperature sensor; 302, Strain feedback unit; 400, Intelligent control center; 401, Multimodal sensing module; 402, Motion planning module; 403, Dynamic optimization module; 500, Multifunctional integrated module; 600, Safety protection module; 601, Dielectric elastomer insulating layer; 602, Redundant emergency stop circuit; 603, Wireless communication module. Detailed Implementation

[0048] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0049] Example 1

[0050] like Figure 1 As shown, an intelligent bathing robot based on multimodal perception and memory alloy drive includes a multi-degree-of-freedom collaborative cleaning system, an intelligent control center 400, and a multi-functional integrated module 500. The multi-degree-of-freedom collaborative cleaning system includes a flexible scrubbing arm 100, a spray arm 200, and a composite joint 300.

[0051] like Figure 1 and Figure 2 As shown, the flexible scrubbing arm 100 and the spray arm 200 are connected by a composite joint 300. The composite joint 300 has a built-in distributed temperature sensor 301 and a strain feedback unit 302 for dynamically compensating for motion errors. The end of the flexible scrubbing arm 100 integrates a MEMS pressure sensor with a range of 0-20N and an accuracy of ±0.1N for real-time monitoring of the scrubbing contact force.

[0052] like Figure 5 As shown, the driving structure of the flexible scrubbing arm 100 includes a flexible skeleton 101, a shape memory alloy wire array 102, and a conical spiral return spring 103. The shape memory alloy wire array 102 consists of groups of nickel-titanium alloy wires spirally wound around the surface of the flexible skeleton. Each group of wires is independently connected to a PID temperature control circuit with a temperature control resolution of ±0.5℃. The conical spiral return spring 103 and the shape memory alloy wire array 102 form an antagonistic driving unit. The elastic modulus of the conical spiral return spring is 1.2-2.5 GPa, the preload is adjustable from 5 to 20 N, the motion frequency of the flexible scrubbing arm 100 is 0.1-5 Hz, the maximum bending angle is ±120°, and the end repeatability is ≤±0.5 mm.

[0053] The spray arm 200 is controlled by a shape memory alloy wire drive unit. The shape memory alloy wire is made of nickel-titanium alloy, with a diameter of 0.3-0.5mm and a phase transition temperature of 35-45℃. Adaptive adjustment of height and angle is achieved through current pulse modulation. A rotatable nozzle is located at the end of the spray arm 200, supporting wide-area spray coverage. The collaborative control method of the spray arm 200 includes: constructing a 3D human body model based on LiDAR point cloud data; calculating the coverage area of ​​each spray arm 200 using a spatial grid partitioning algorithm; allocating spray tasks using a game theory optimization model, with the optimization objectives of water flow uniformity and energy minimization, dynamically adjusting the height and angle of the spray arm 200; and controlling the spray flow rate via pulse width modulation (PWM), with a flow rate adjustment range of 0.5-5L / min and a response time ≤50ms.

[0054] like Figure 3As shown, the intelligent control hub 400, developed based on the ROS architecture, includes a multimodal perception module 401, a motion planning module 402, and a dynamic optimization module 403. The multimodal perception module 401 integrates a binocular vision unit, a 16-line LiDAR, and a voice interaction module. The binocular vision unit has a positioning accuracy of ±1.5mm, and the LiDAR has an angular resolution of 0.1°. The voice interaction module's functions include: supporting Mandarin Chinese and dialect recognition; a voice command set containing 20 types of control commands with a recognition accuracy ≥95%; integrating a voiceprint recognition unit, which can bind up to 5 user identities and invoke personalized bathing programs based on identity information; voice feedback latency ≤300ms; and using TTS technology to generate natural language prompts. The motion planning module 402 uses an improved RRT algorithm to generate a three-dimensional curved surface scrubbing trajectory. The improved RRT algorithm introduces human curvature constraints, resulting in a trajectory tracking error ≤±2mm. The optimization steps of the improved RRT algorithm include: introducing human body surface curvature constraints into the standard RRT algorithm, and generating a smooth scrubbing path through B-spline curve fitting; adopting a dynamic step size adjustment strategy, with an initial step size of 10mm, and adaptively reducing the step size to 2mm when the detected curvature radius is <50mm; adding a contact force weight factor in the path optimization stage to ensure real-time coupling between trajectory planning and pressure feedback. The dynamic optimization module 403 adjusts the scrubbing force, spray intensity, and massage parameters in real time through a fuzzy PID controller. The fuzzy PID controller takes contact force deviation, posture deviation, and water flow pressure as input variables, and outputs drive current, spray flow rate, and vibration frequency. The implementation method of the fuzzy PID controller includes: defining the contact force deviation e(t), the deviation change rate ec(t), and the water flow pressure p(t) as input variables, with universes of discourse of [-5N, 5N], [-2N / s, 2N / s], and [0.1MPa, 0.5MPa], respectively; the output variables are the shape memory alloy driving current I(t), the spray valve opening α(t), and the vibration frequency f(t), with universes of discourse of [0.1A, 2A], [0%, 100%], and [10Hz, 100Hz], respectively; the fuzzy rule base contains 81 IF-THEN rules, and the centroid method is used to defuzzify the rules, with a control cycle ≤10ms.

[0055] The multi-functional integrated module 500 includes a hot air drying unit, an embedded massage mechanism, and a data fusion unit. The hot air drying unit features an infrared thermometer for real-time monitoring of skin surface temperature, a fan and heating structure to generate hot air, and a temperature controller. The infrared thermometer, fan, and heating structure are all electrically connected to the temperature controller. The outlet temperature and airflow of the hot air drying unit are adjustable, with an adjustable temperature range of 40-60℃ and an airflow range of 0.5-3 m / s. A PID temperature control algorithm maintains temperature fluctuations ≤ ±1℃. The hot air drying unit operates in two modes: a rapid drying mode and a comfortable drying mode. In the rapid drying mode, the temperature is set at 60℃ and the airflow at 3 m / s for 5-10 minutes; in the comfortable drying mode, the temperature is set at 45℃ and the airflow at 1.5 m / s for 15-20 minutes. Temperature control employs a feedforward-feedback composite control strategy, using an infrared thermometer to monitor skin surface temperature in real-time, with an overshoot of ≤1℃.

[0056] The embedded massage mechanism features a built-in eccentric wheel vibration mechanism with a continuously adjustable vibration frequency of 10-100Hz and an amplitude of 0.1-5mm, supporting switching between dot matrix and waveform massage modes. The control logic of the embedded massage mechanism includes: automatically matching the massage mode based on the contact force distribution feedback from the MEMS pressure sensor: switching to a low-frequency deep pressure mode (10-30Hz) when the pressure > 5N, and switching to a high-frequency kneading mode (50-100Hz) when the pressure ≤ 5N; the massage path is generated based on an improved RRT* algorithm and alternates with wiping actions, with an adjustable time interval of 5-30 seconds. The data fusion unit integrates multi-sensor data through a Kalman filter algorithm to achieve coordinated control of bathing, massage, and care actions. The workflow of the data fusion unit includes: synchronizing binocular vision, lidar, and inertial measurement unit (IMU) data via timestamps; fusing multi-source information using the extended Kalman filter (EKF) algorithm to output a 6-DOF pose estimate with an attitude angle error ≤ ±0.5°; and triggering an emergency braking protocol when a sudden human motion acceleration > 2 m / s² is detected, with a response delay ≤ 100 ms.

[0057] like Figure 4 As shown, the present invention also includes a safety protection module 600, which includes: a dielectric elastomer insulating layer 601 covering the surface of the flexible scrubbing arm 100, with a breakdown voltage ≥5kV; and a redundant emergency stop circuit 602 that cuts off the drive power and activates mechanical self-locking when the contact force exceeds the limit >15N or the temperature is abnormal >50℃. Furthermore, it also includes a wireless communication module 603, which can upload the device status to a monitoring terminal in real time, supporting remote emergency intervention.

[0058] Example 2

[0059] This embodiment provides a compliance control for a memory alloy driven scrubbing arm, and the specific steps are as follows:

[0060] 1) The user gives the voice command "Start cleaning". The voice interaction module recognizes the command and sends it to the intelligent control center 400.

[0061] 2) The binocular vision unit scans the curved surface of the user's back and generates a 3D point cloud model with an accuracy of ±1.5mm;

[0062] 3) The shape memory alloy drive module receives current pulses, the nickel-titanium alloy wire is heated and shrinks, driving the flexible scrubbing arm 100 to reciprocate at a frequency of 2Hz.

[0063] 4) The MEMS pressure sensor monitors the contact force in real time. If it exceeds 8N, the fuzzy PID controller automatically reduces the drive current to 0.3A to ensure that the force fluctuation is ≤±0.2N.

[0064] Example 3

[0065] This embodiment provides a method for large-area spray coverage, the specific steps of which are as follows:

[0066] 1) LiDAR is used to construct a bathroom environment model and mark key areas of the user's shoulders and waist;

[0067] 2) The motion planning module 402 plans the spray path through an improved RRT algorithm and dynamically adjusts the spray arm elevation angle from 200 to 45° with a spacing of 300mm;

[0068] 3) When the shape memory alloy wire is energized, it contracts, causing the nozzle to rotate 30°, expanding the water flow coverage to a fan shape of 120°;

[0069] 4) The flow sensor feeds back data to the control center to maintain a total flow rate of 3L / min, a single-arm flow rate of 1.5L / min, and a constant water temperature of 38℃±1℃.

[0070] Example 4

[0071] This embodiment provides a hot air drying and massage combined mode, which specifically includes the following steps:

[0072] 1) After showering, the hot air drying unit starts PID temperature control, the air outlet temperature rises to 50℃, and the wind speed is 2m / s;

[0073] 2) The embedded massage mechanism switches to high-frequency kneading mode, 60Hz, amplitude 2mm, and moves along the spinal area for massage;

[0074] 3) The infrared thermometer monitors the skin surface temperature. If it exceeds 42°C, it automatically lowers the air temperature to 45°C and increases the massage interval to 10 seconds.

[0075] 4) After drying is complete, the voice module announces "Nursing is complete" and uploads the nursing log to the cloud.

[0076] Example 5

[0077] This embodiment provides an emergency safety braking response, which specifically includes the following steps:

[0078] 1) The inertial measurement unit detects abnormal acceleration of the robotic arm (>3m / s²) and triggers a level one alarm;

[0079] 2) The redundant emergency stop circuit 602 cuts off the memory alloy drive power supply, and the conical return spring forces the joint to return to its original position;

[0080] 3) The dielectric elastomer insulation layer 601 starts a self-test. If the insulation resistance is <10MΩ, a fault code is sent to the monitoring terminal through the wireless module 603.

[0081] 4) The system is locked to standby mode and requires administrator fingerprint authentication before restarting.

[0082] Example 6

[0083] This embodiment provides a method for invoking a personalized bathing plan, which specifically includes the following steps:

[0084] 1) After the user's voiceprint recognition is successful, the intelligent control center 400 loads the preset "semi-reclining bathing mode";

[0085] 2) The posture adjustment mechanism adjusts the bathing seat to a 55° tilt angle, and the flexible scrubbing arm 100 cleans the upper limbs according to the preset trajectory, with a frequency of 1Hz and a force of 5N±0.3N;

[0086] 3) The spray arm 200 is switched to low flow mode, 0.8L / min, and the water temperature is increased to 40℃;

[0087] 4) The massage mechanism synchronously starts low-frequency vibration and automatically stores user preference parameters after the bath.

[0088] In summary, this invention solves the problems of insufficient compliance, poor environmental adaptability, high energy consumption, and limited functionality of traditional bathing robots through biomimetic drive design, multi-source perception fusion, and closed-loop control optimization. It achieves technological breakthroughs in safety, intelligence, and user experience, providing an efficient and comfortable integrated solution for the field of elderly care and rehabilitation aids.

[0089] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A multi-modal perception and memory alloy driven intelligent bathing robot, characterized in that, include: A multi-degree-of-freedom collaborative cleaning system, specifically including: A flexible scrubbing arm with an integrated MEMS pressure sensor at its end for real-time monitoring of scrubbing contact force; The spray arm, controlled by a shape memory alloy wire drive unit, can achieve adaptive adjustment of height and angle through current pulse modulation. The end of the spray arm is equipped with a rotatable nozzle. The flexible scrubbing arm and the spray arm are connected by a composite joint, which has a built-in distributed temperature sensor and a strain feedback unit for dynamically compensating for motion errors. The intelligent control center specifically includes: The multimodal perception module integrates a binocular vision unit, LiDAR, and a voice interaction module; The motion planning module is used to generate scrubbing trajectories that adapt to the curved surfaces of the human body; The dynamic optimization module is used to adjust the scrubbing intensity, spray strength, and massage parameters in real time. The multi-functional integrated module specifically includes: Hot air drying unit for human body drying; Embedded massage mechanism for human body massage; The data fusion unit integrates data from multiple sensors using a Kalman filter algorithm to achieve coordinated control of bathing, massage, and nursing actions.

2. The multi-modal perception and memory alloy driven smart bathing robot according to claim 1, wherein, The drive structure of the flexible scrubbing arm includes: A shape memory alloy wire array consisting of multiple sets of nickel-titanium alloy wires, wherein the nickel-titanium alloy wires are spirally wound on the surface of a flexible skeleton and each is connected to an independent temperature control circuit. A conical helical return spring, which together with the memory alloy wire array forms an antagonistic drive unit.

3. The multi-modal perception and memory alloy driven smart bathing robot of claim 1, wherein, The coordinated control method for the spray arm includes: A three-dimensional human body model is constructed based on the sensing data from the multimodal sensing module, and the coverage area of ​​each spray arm is calculated. A game theory optimization model is used to allocate spraying tasks and adjust the posture of the spraying arms; And the spray flow rate is controlled by pulse width modulation.

4. The multi-modal perception and memory alloy driven smart bathing robot of claim 1, wherein, The motion planning module employs an improved RRT algorithm and incorporates human curvature constraints to generate a scrubbing trajectory adapted to the human body's surface. Specific steps include: Introduce human body surface curvature constraints into the standard RRT algorithm, and generate a smooth scrubbing path by fitting B-spline curves. A step size strategy that is adaptively adjusted based on the radius of curvature is adopted; In the path optimization stage, a contact force weighting factor is added to achieve real-time coupling of trajectory planning and pressure feedback.

5. The multi-modal perception and memory alloy driven smart bathing robot of claim 1, wherein, The dynamic optimization module adjusts the scrubbing force, spray intensity and massage parameters in real time through a fuzzy PID controller. The fuzzy PID controller takes contact force deviation, posture deviation and water flow pressure as input variables and outputs drive current, spray flow rate and vibration frequency. The fuzzy PID controller uses contact force deviation, deviation change rate and water flow pressure as input variables, and shape memory alloy drive current, spray valve opening degree and vibration frequency as output variables, and includes a preset fuzzy rule base.

6. The multi-modal perception and memory alloy driven smart bathing robot of claim 1, wherein, The hot air drying unit includes an infrared thermometer for real-time monitoring of skin surface temperature, a fan and heating structure for generating hot air, and a temperature controller. The infrared thermometer, fan, and heating structure are all electrically connected to the temperature controller. The hot air drying unit adopts a feedforward-feedback composite control strategy and is equipped with multiple drying modes, including a fast drying mode and a comfortable drying mode. The set temperature of the fast drying mode is higher than that of the comfortable drying mode, and the airflow speed of the fast drying mode is higher than that of the comfortable drying mode.

7. The multi-modal perception and memory alloy driven smart bathing robot of claim 1, wherein, The embedded massage mechanism can execute a low-frequency deep pressure mode and a high-frequency kneading mode, and automatically matches the massage mode according to the contact force distribution fed back by the MEMS pressure sensor: when the pressure is greater than the set pressure threshold, it switches to the low-frequency deep pressure mode, and when the pressure is less than or equal to the set pressure threshold, it switches to the high-frequency kneading mode; the frequency range of the low-frequency deep pressure mode is 10-30Hz, and the frequency range of the high-frequency kneading mode is 50-100Hz; the massage path of the embedded massage mechanism is generated based on an improved RRT algorithm and is executed alternately with the wiping action, and the time interval can be set.

8. The multi-modal perception and memory alloy driven smart bathing robot of claim 1, wherein, An inertial measurement unit is provided inside the composite joint or at the end of the flexible scrubbing arm. The workflow of the data fusion unit includes: Synchronize binocular vision, lidar, and inertial measurement unit data using timestamps; An extended Kalman filter algorithm is used to fuse information from multiple sources and output a 6-DOF pose estimate. When a sudden human movement is detected, an emergency braking protocol is triggered. The sudden movement refers to the absolute value of acceleration collected in real time by the inertial measurement unit exceeding the set acceleration threshold.

9. The multi-modal perception and memory alloy driven smart bathing robot of claim 1, wherein, The voice interaction module integrates a voice command recognition unit and a voiceprint recognition unit. The voice command recognition unit has a voice command set, which includes multiple control commands. The voiceprint recognition unit can bind multiple users, and each user has a corresponding personalized bathing plan. The personalized bathing plan includes, but is not limited to, scrubbing contact force, spray water temperature and flow rate, drying mode and massage mode. 10.The intelligent bathing robot based on multi-modal perception and memory alloy driving according to any one of claims 1-9, characterized in that, It also includes a safety protection module, which includes a dielectric elastomer insulating layer covering the surface of the flexible scrubbing arm, a redundant emergency stop circuit for cutting off the power and activating mechanical self-locking under abnormal conditions, and a wireless communication module for status uploading and remote intervention; the breakdown voltage of the dielectric elastomer insulating layer is ≥5kV; the abnormal conditions include when the contact force exceeds the set warning pressure threshold and when the temperature exceeds the set abnormal temperature threshold.

Images