Healthy old-age care robot system based on bimodal mobile chassis and control method

Abstract

According to the healthy old-age care service robot based on the bimodal moving chassis, the operation stability of a mechanical arm is remarkably improved through intelligent switching of a moving form (omnidirectional wheel / crawler driving) and a fixed form (four-corner supporting legs and bionic suction cups). The system integrates a multi-degree-of-freedom mechanical arm, a multi-level sensor and a light-weight AI algorithm, core functions such as tea and water supply, medicine and object feeding and health monitoring are achieved, and the cost of the whole machine is reduced to 30% of that of a traditional scheme. Innovatives include a bimodal chassis control algorithm, a magnetic suction quick-change mechanical arm interface, millimeter wave radar respiration detection and the like, and large-scale mass production feasibility is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of service robots, and particularly to a low-cost elderly care service robot integrating a dual-modal chassis, a multi-degree-of-freedom robotic arm, and a health monitoring system, which is applicable to the elderly care scenarios in families and elderly care institutions. Background Art

[0002] At present, the pain points in the industry of health elderly care service robots are as follows: (1) The existing service robots have insufficient mobile stability, and the robotic arm is prone to jitter during operation; (2) Lack of a dynamic stability control mechanism for the behavioral characteristics of the elderly; (3) The traditional support system has a complex structure and high cost.

[0003] References to the prior art: (1) CN2025102787804 A Support and Protection System for a Dual-Modal Mobile Robot with Fixed and Expandable Leg Numbers (by the same applicant): Discloses a support and protection system for a dual-modal mobile robot with fixed and expandable leg numbers, including two core forms: the form with a fixed number of legs, pre-installed with one of a triangle / quadrangle / pentagon / hexagon at the factory, using an integrated molding process to reduce costs, and applicable to standardized scenarios; the form with an expandable number of legs, dynamically increasing or decreasing the legs (3 - 6) through standardized interfaces, and adapting to heavy loads and complex terrains; (2) CN2025103067158 A Multi-Radar Vital Sign Monitoring System Applied to Fixed Population Places (by the same applicant): Discloses a multi-radar vital sign monitoring system applied to fixed population places, including a horizontal layer of 24GHz radar network, a vertical layer of 60GHz radar array, and an auxiliary sensor module. Through the three-dimensional collaborative deployment of heterogeneous radars, combined with an improved variational mode decomposition algorithm and a joint probabilistic data association tracking technology, the separation of respiration, heart rate, and behavioral characteristics of multiple targets in a dense scenario is achieved. Summary of the Invention

[0004] The present invention relates to a health elderly care service robot based on a dual-modal mobile chassis, which significantly improves the operation stability of the robotic arm through the intelligent switching between a mobile form (omnidirectional wheel / crawler drive) and a fixed form (quadrilateral support legs + bionic suction cups). The system integrates a multi-degree-of-freedom robotic arm, multi-level sensors, and a lightweight AI algorithm to realize core functions such as serving tea and water, feeding medicine and delivering items, and health monitoring.

[0005] The hardware system of the health care and elderly care service robot based on a dual-mode mobile chassis is a three-dimensional structure including an omnidirectional wheel chassis (diameter 400 mm), a vertical lifting column (stroke 0.3 - 1.2 m), dual robotic arms (6 degrees of freedom for each arm), a top circular tray (diameter 250 mm including RFID tag area), and sensor layout ( Figure 1 ). The innovation lies in: the mobile chassis adopts a dual-mode structure: a retractable four-corner support mechanism is added to the bottom of the chassis. Mobile form: omnidirectional wheels / caterpillars contact the ground; fixed form: the four-corner support legs are fully extended (leg length 200 mm), and the end bionic suction cups adsorb the ground, and an expandable leg system (3 - 6 legs with dynamic configuration), inheriting the standardized interface design from CN2025102787804.

[0006] The working process of the dual-mode chassis is as follows: Step 1, when the load of the robotic arm > 2 kg, trigger the deployment of the support legs; Step 2, the suction cups pre-press the ground, and the micro vacuum pump pumps air to -80 kPa; Step 3, the robotic arm performs operations (end jitter < 0.3 mm).

[0007] Privacy protection design: a physical shielding device for the camera (the sliding cover opening and closing state is visible). Usually, the camera sliding cover is closed to physically shield and protect privacy, and in an emergency, the sliding cover automatically opens to provide on-site video.

[0008] Waterproof treatment: silicone rubber sealing strips at the body joints (IPX4 level).

[0009] Figure 2 It is the mechanical linkage schematic diagram for the dual-mode drive of the mobile chassis. The omnidirectional wheels use a planetary gear set (reduction ratio 15:1) to achieve 360° translation; the caterpillar module is connected to the chassis through an electromagnetic latch, and the ground clearance is increased to 50 mm after deployment; the power system shares the same brushless motor (peak torque 12 N·m), and the mode is switched through a clutch.

[0010] Figure 3 It is the distribution diagram of the degrees of freedom of the robotic arm joints. The movement ranges of the joints of the dual robotic arms are: base rotation joint (±180°), shoulder pitch joint (-30° to 90°), elbow telescopic joint (stroke 200 mm), and wrist three-degree-of-freedom flexible joint (including ±45° deflection). The innovative design lies in the lightweight structure using carbon fiber connecting rods + 3D printed joints (self-weight of each arm < 1.5 kg), and the gripper / suction cup / bionic hand can be quickly replaced through a magnetic quick-change interface.

[0011] This robot adopts the coordination of a multi-level perception system: (1) Navigation layer: A 16-line lidar at the top (detection radius 8m) and an ultrasonic array on the chassis (covering 6 directions), a 24GHz millimeter-wave radar (with strong anti-attenuation performance, used for long-distance breathing monitoring); (2) Operation layer: A 60GHz near-field radar for the robotic arm (with high resolution, used for lip positioning and tracking of the medicine-feeding action); (3) Health monitoring layer: An infrared temperature sensor array under the tray (5×5 dot matrix, accuracy ±0.2°C); (4) Privacy and security layer: A physical shielding state sensor for the camera (using a Hall element to detect the position of the sliding cover), and a 77GHz short-pulse radar can be used in combination when necessary (penetrating non-metallic materials, used for non-contact heart rate detection).

[0012] In the perception system algorithm, the Joint Variational Mode Decomposition (JVMD) algorithm is introduced to separate multi-target life signals in dense scenarios (accuracy improved by 32%). A distributed data processing architecture based on federated learning is used to reduce the dependence on the cloud (the proportion of edge-side computing > 70%).

[0013] The software system of this robot ( Figure 4 ) adopts a hierarchical architecture, and its core modules include: (1) Perception layer: Multi-sensor data fusion (lidar + visual SLAM positioning error < 3cm); (2) Decision-making layer: A lightweight object recognition model (YOLO-FastestV2, 10fps@Jetson Nano); (3) Control layer: Cooperative motion planning for dual robotic arms (collision avoidance algorithm based on Lie groups and Lie algebras); (4) Interaction layer: A voice command parsing system (supporting fuzzy semantic matching, such as "Get that round box"). Description of the Drawings

[0014] Figure 1 : Three-dimensional schematic diagram of the overall machine structure (modular design expressed in the form of a class diagram).

[0015] Figure 2 : Detailed drawing of the mobile chassis transmission mechanism (flowchart expressing mode switching).

[0016] Figure 3 : Distribution diagram of the degrees of freedom of the robotic arm joints (motion chain expressed in a tree structure).

[0017] Figure 4 : Software system architecture diagram (layered data flow).

[0018] Figure 5 : Flowchart of typical working scenarios (sequence diagram with conditional judgment). Specific Implementation Modes

[0019] Taking the "feeding medicine to the mouth" task as an example to illustrate the whole process Figure 5 ): ① Voice receiving instruction → ② Medicine positioning (tray RFID matching + visual verification) → ③ Robotic arm picking up medicine (force feedback control, grasping force ≤ 2N) → ④ Navigating to the old person's position (UWB following mode) → ⑤ Mouth positioning (lip recognition by close-up camera) → ⑥ Delivery action (joint impedance control to prevent collision) → ⑦ Task completion confirmation (pressure sensor feedback for pill detachment).

[0020] The implementation method of health monitoring and care assistance includes the standardized process of item delivery, the implementation method of health monitoring, and behavior analysis.

[0021] Standardized process of item delivery (the key point protected by the claim): Step 1: Item positioning. High-frequency items are pre-stored in the tray RFID area (medicine boxes are embedded with passive tags); dynamic items are identified through vision (the training set contains 300° perspective images of 20 types of commonly used items for the elderly). Step 2: Safe grasping. Suction cup mode: for items with smooth surfaces (such as water cups, vacuum degree ≥ 0.6 bar); three-finger gripper or five-finger dexterous hand: suitable for irregular items (medicine bottles, adaptive envelope grasping). Step 3: Delivery control. Mouth positioning: the close-up camera captures lip features (error < 5mm); anti-shake algorithm: IMU data fusion control at the end of the robotic arm (vibration suppression > 70%).

[0022] The implementation method of health monitoring uses non-contact vital sign detection: infrared array measures the body surface temperature distribution (abnormal fever warning), millimeter-wave radar monitors the breathing frequency (0.5 - 2Hz, accuracy ± 0.1Hz), and when adding a 77GHz short-pulse radar (non-contact heart rate monitoring, penetrating non-metal obstacles), the heart rate can be monitored.

[0023] Behavior analysis: The skeleton point tracking algorithm (optimized version of OpenPose) detects falls (sensitivity 93%); Medication compliance monitoring: jointly judged by the weight sensor of the medicine box and the medicine-taking action.

Claims

1. A health care and elderly care service robot with a dual-modal mobile chassis, characterized in that include: (a) Dual-mode mobile chassis (mobility mode: omnidirectional wheel / track drive; Fixed form: expandable leg system + bionic suction cup adsorption); (b) Liftable dual-arm system (single arm with 6 degrees of freedom, with magnetic quick-change interface at the end); (c) Multi-layer sensor array (navigation layer, operation layer, health monitoring layer); (d) Multi-band radar vital signs monitoring system (24GHz / 60GHz / 77GHz working together).

2. The robot according to claim 1, characterized in that The fixed form includes: (1) Scalable outrigger system (3-6 outriggers dynamically configured, inherited from the standardized interface of CN2025102787804); (2) Bionic suction cup module at the end of the leg (dynamic adjustment range of suction force -30 kPa to -80 kPa); (3) Trigger conditions: Automatic switching when the robot arm load is >2kg or the ground vibration is >0.5g.

3. The robotic arm system according to claim 1(b), characterized in that: (1) Single arm with 6 degrees of freedom joint (base rotation ±180°, shoulder pitch -30°~90°, elbow extension 200mm, wrist with 3 degrees of freedom flexible joint); (2) The end effector includes a close-up camera (lip positioning accuracy ±5 mm) and a force feedback sensor (grasping force control range 1-5 N).

4. The sensor array according to claim 1 (c), characterized in that Include: (1) Navigation layer: 16-line laser radar (detection radius 8m) + ultrasonic array (coverage in 6 directions); (2) Health monitoring layer: 24GHz millimeter wave radar (breathing detection) + 60GHz near-field radar (heart rate detection); (3) Privacy protection layer: 77 GHz short pulse radar (non-contact heartbeat monitoring, penetrating non-metallic obstructions).

5. The vital signs monitoring system according to claim 1(d), characterized in that: (11) Using the joint variational mode decomposition (JVMD) algorithm to separate multi-target life signals; (2) Dynamic frequency band switching mechanism (when 60 GHz radar sampling is poor, 24 GHz sampling frequency is enhanced); (3) The heterogeneous radar 3D deployment architecture inherited from CN2025103067158.

6. The robot according to claim 1, wherein The software system includes: (1) Dual-mode switching control algorithm (response time <1.5 seconds, including synchronous locking of chassis power when outriggers are deployed); (2) Item delivery decision engine (YOLO-FastestV2 model + dynamic obstacle avoidance path planning); (3) Distributed data processing under the federated learning framework (end-side computing accounts for > 70%).

7. The robot according to claim 1, wherein Waterproof design: (1) The whole machine has an IPX4 protection level (the circuit board has a three-conformal paint spray thickness of 50-80μm); (2) Labyrinth sealing structure of the robot arm joint (silicone sealing ring + guide groove); (3) Suction cup module drainage channel (aperture 0.5 mm, inclination angle 5°).

8. The robot according to claim 1, characterized in that Working Mode: (1) Close following mode (UWB positioning accuracy ±10cm, maximum following speed 0.8m / s); (2) Emergency rescue mode (fall detection sensitivity 93%, automatically dialing preset contacts and sending location).

9. The outrigger system according to claim 2, wherein the control method thereof comprises: (1) Floor material recognition (classified into 5 types based on lidar reflection intensity); (2) Adsorption force adaptive adjustment (70 kPa for smooth floor tiles, 30 kPa for felt carpets); (3) Disengagement safety protocol (releasing negative pressure after 3 seconds of audible and visual warning, and the support legs retracting in stages).

Images