Upper limb exoskeleton rehabilitation robot with charging and discharging functions
By introducing a reversible motor-generator actuator and a dual-agent adaptive framework into the upper limb exoskeleton rehabilitation robot, the problem of the separation between energy recovery and training mode is solved, achieving efficient energy utilization and human-machine collaboration, and improving the portability of the device and the training effect.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-30
- Publication Date
- 2026-04-14
AI Technical Summary
Existing upper limb exoskeleton rehabilitation robots have a single energy flow direction, and the mechanical energy generated by the patient's active movement is not effectively recovered and utilized. The training mode is disconnected from energy management, and the human-machine collaboration capability is insufficient, resulting in poor equipment portability, low training continuity, and poor comfort.
Design an upper limb exoskeleton rehabilitation robot with charging and discharging functions. It adopts a reversible motor-generator actuator, which generates electrical energy through patient movement, stores it, and provides assistance in active training mode. Combined with sensors and control modules, it realizes real-time matching of energy flow and training mode. Human-computer interaction is optimized based on a dual-agent mutual adaptation framework.
It enables continuous operation of the equipment in scenarios without external power supply, improves energy utilization efficiency and training continuity, and enhances human-machine collaboration efficiency and comfort.
Smart Images
Figure CN121845903A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the fields of rehabilitation engineering and robotics, and in particular to an upper limb exoskeleton rehabilitation robot with charging and discharging functions. Background Technology
[0002] With the ongoing aging population and the rising incidence of neurological diseases such as stroke and spinal cord injury, the number of patients with upper limb motor dysfunction is increasing, making the demand for professional rehabilitation training increasingly urgent. Upper limb exoskeleton rehabilitation robots, as an important product at the intersection of rehabilitation engineering and robotics technology, can provide patients with standardized and personalized rehabilitation training support, helping them regain upper limb motor abilities. They have become one of the key devices for clinical and home-based rehabilitation.
[0003] Existing upper limb exoskeleton rehabilitation robots mostly employ a unidirectional energy transfer mode, where an external power source powers the drive motor, which outputs torque to move the mechanical joints, thereby enabling the patient's upper limb to complete rehabilitation training. During training, the mechanical energy generated by the patient's upper limb movements is typically dissipated as heat through friction and damping, without effective recovery and utilization. Energy management and training modes are independent of each other, lacking an integrated design. The training process relies entirely on continuous external power supply, and the training modes are mostly preset with fixed parameters, allowing only limited adjustments based on simple sensor data.
[0004] In summary, the shortcomings of existing technologies are as follows: First, existing technologies suffer from a single energy flow path, resulting in the waste of mechanical energy generated by the patient's active movements. This reduces energy utilization efficiency and increases reliance on external power sources, making continuous operation difficult in situations without external power supply and limiting the device's portability and application scenarios. Second, training modes and energy management are disconnected, failing to organically integrate energy recovery with rehabilitation training needs. The system cannot dynamically optimize energy allocation based on the patient's movement status, impacting the continuity and cost-effectiveness of training. Third, human-machine collaboration and adaptability are insufficient. Existing control strategies often employ fixed control methods, making it difficult to respond in real-time to changes in the patient's movement intentions, differences in muscle activation states, and functional improvements during rehabilitation. This leads to poor naturalness and comfort in assisted output and low efficiency in human-machine collaboration. Based on these issues, there is an urgent need to design a novel upper limb exoskeleton rehabilitation robot that integrates energy recovery, intelligent charging and discharging, and efficient human-machine collaboration. Summary of the Invention
[0005] To address the shortcomings of existing technologies, this invention provides an upper limb exoskeleton rehabilitation robot with charging and discharging functions. In passive training mode, this invention can convert and store the mechanical energy generated by the patient's upper limb movements, enabling simultaneous training and charging. In active training mode, the stored electrical energy is used to power the joint drive module, providing assistance or guiding movement for the patient, thus forming an upper limb exoskeleton rehabilitation robot that integrates passive training, energy recovery, and active assistance.
[0006] To achieve the above objectives, the present invention adopts the following technical solution: An upper limb exoskeleton rehabilitation robot with charging and discharging functions includes: The system comprises a mechanical body module, multiple joint drive modules, an energy storage module, a control module, and a sensor module. The mechanical body module is fixedly connected to the patient's upper limb to be rehabilitated. The joint drive modules have dual operating modes: a motor and a generator. In generator mode, the module generates electrical energy with the movement of the patient's upper limb and charges the energy storage module. In motor mode, the energy storage module powers the movement of the patient's upper limb. The energy storage module is electrically connected to the joint drive module, and the control module is electrically connected to both the joint drive module and the energy storage module. The sensor module is used to collect motion information of the patient's upper limb to be rehabilitated and feed it back to the control module.
[0007] As a further technical solution, the joint drive module is installed on the mechanical body module, which includes at least a shoulder joint unit and an elbow joint unit, and each joint drive module is installed on the shoulder joint unit and the elbow joint unit respectively.
[0008] As a further technical solution, the joint drive module includes a reversible motor-generator actuator and a reduction transmission mechanism connected to the reversible motor-generator actuator. The reversible motor-generator actuator adopts a permanent magnet synchronous motor and can switch between motor working mode and generator working mode.
[0009] As a further technical solution, the energy storage module includes a rechargeable battery and a supercapacitor, used to store electrical energy generated in the generator operating mode and to supply power to the joint drive module in the motor operating mode.
[0010] As a further technical solution, each joint drive module is connected to the energy storage module via a DC bus; the control module obtains the voltage of the DC bus and realizes unified scheduling of energy between multiple joints based on the voltage of the DC bus.
[0011] As a further technical solution, the sensor module includes an inertial measurement unit mounted on the mechanical body module and corresponding to the patient's upper arm or forearm. The inertial measurement unit is used to measure upper limb posture data.
[0012] As a further technical solution, the control module acquires upper limb posture data through an inertial measurement unit and combines it with joint angle data. In passive training mode, it adjusts the power damping of the joint drive module according to the posture data and joint angle data. In active training mode, it adjusts the assist torque of the joint drive module in real time according to the posture data, joint angle data, and electromyographic signals.
[0013] As a further technical solution, the sensor module also includes an electromyography (EMG) sensor attached to the relevant muscle group of the patient's upper limb. The EMG sensor is used to detect the degree of muscle activation in the patient and output EMG signals.
[0014] As a further technical solution, the control module is based on a dual-agent adaptive framework, specifically: the patient is regarded as a human intelligent agent, and the upper limb exoskeleton rehabilitation robot is regarded as a machine intelligent agent. Through multi-model control and reinforcement learning algorithms, the control parameters and training modes are updated online according to motion deviation, muscle load and comfort indicators.
[0015] As a further technical solution, the control module includes multiple training modes, specifically including passive charging mode, damping training mode, semi-active assist mode and active assist mode, which can be selected and switched through a human-computer interaction interface.
[0016] The present invention has the following beneficial effects: (1) This invention incorporates a reversible motor-generator actuator with both motor and generator operating modes within the joint drive module. Combined with the charging and discharging function of the energy storage module, in passive training mode, the patient's active movement drives the reversible motor-generator actuator to switch to generator operating mode, storing the generated electrical energy in the energy storage module. In active training mode, the energy storage module directly supplies power to the reversible motor-generator actuator, causing it to switch to motor operating mode to assist or guide the patient's movement. This fully utilizes the energy generated by the patient's own movement, reducing mechanical energy waste and significantly lowering the device's dependence on external power. Even in scenarios without external power, active training can be maintained using the recovered energy.
[0017] (2) The control module of this invention is electrically connected to the joint drive module and the energy storage module, deeply integrating the rehabilitation training process with the energy management process, thus solving the problem of the two being separated in the prior art. The control module controls the switching between motor mode and generator mode based on the patient's upper limb movement information collected by the sensor module, and simultaneously regulates the charging and discharging state of the energy storage module, so that the energy flow matches the training needs in real time, ensuring the continuity and smoothness of rehabilitation training and improving training efficiency.
[0018] (3) This invention collects the patient's upper limb movement information in real time through the sensor module and feeds it back to the control module, providing the control module with accurate human-computer interaction status data support. The control module realizes online learning and adaptive optimization of the human-computer interaction status based on the dual-agent mutual adaptation framework, improving the naturalness, comfort and human-computer collaboration efficiency of the assistance. Attached Figure Description
[0019] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.
[0020] Figure 1 A schematic diagram of the overall structure of an upper limb exoskeleton rehabilitation robot with charging and discharging functions provided by the present invention; Figure 2 This is a schematic diagram of the electromechanical coupling structure and signal flow of the joint drive module in this invention; Figure 3 This is a schematic diagram illustrating the conversion and flow direction between mechanical energy and electrical energy under two different working modes of the present invention; Figure 4 This is a strategy framework diagram of the control module of the present invention; Figure 5 This invention provides a human-machine collaborative framework based on control theory. Detailed Implementation
[0021] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.
[0022] Example 1 like Figure 1 As shown, this invention provides an upper limb exoskeleton rehabilitation robot with charging and discharging functions, belonging to the field of rehabilitation engineering and robotics technology, specifically including: The system includes a mechanical body module, multiple joint drive modules, an energy storage module, a control module, and a sensor module, wherein the control module is electrically connected to the joint drive module and the energy storage module.
[0023] In this embodiment, the mechanical body module is fixedly connected to the patient's upper limb to be rehabilitated, and the mechanical body module includes at least a shoulder joint unit and an elbow joint unit; multiple joint drive modules are installed on the mechanical body module, specifically: the joint drive modules include, but are not limited to, a shoulder joint drive module and an elbow joint drive module, and are correspondingly installed on the shoulder joint unit and the elbow joint unit.
[0024] In this embodiment, as Figure 2 As shown, the joint drive module has dual operating modes: motor and generator. Specifically, the joint drive module includes a reversible motor-generator actuator and a reduction gear transmission mechanism connected to the reversible motor-generator actuator. The reversible motor-generator actuator uses a permanent magnet synchronous motor and can switch between motor operating mode and generator operating mode. Figure 3 As shown, in passive training mode, the reversible motor-generator actuator switches to generator mode, generating electrical energy with the patient's upper limb movement and charging the energy storage module; in active training mode, the reversible motor-generator actuator switches to motor mode, and the energy storage module powers the patient's upper limb movement.
[0025] The control module can adjust the equivalent resistance or current of the reversible motor-generator actuator in passive training mode, so that the joint drive module (not limited to the shoulder joint drive module and the elbow joint drive module) can generate adjustable damping for the patient, realize different levels of resistance training and take into account energy recovery efficiency.
[0026] In this embodiment, the energy storage module is electrically connected to the joint drive module. The energy storage module includes a rechargeable battery and a supercapacitor, which are used to store electrical energy generated in the generator working mode and to supply power to the joint drive module in the motor working mode. The control module manages the energy distribution between the rechargeable battery and the supercapacitor according to the training intensity and the remaining power of the energy storage module.
[0027] In this embodiment, as Figure 1 As shown, each joint drive module is connected to the energy storage module via a DC bus, and the control module obtains the voltage of the DC bus and realizes unified scheduling of energy between multiple joints based on the voltage of the DC bus.
[0028] The sensor module includes an inertial measurement unit (IMU) mounted on the mechanical body module and corresponding to the patient's upper arm or forearm. The inertial measurement unit is used to measure upper limb posture data and joint angle data.
[0029] In this embodiment, the sensor module also includes an electromyography (EMG) sensor attached to the relevant muscle group of the patient's upper limb. The EMG sensor is used to detect the degree of muscle activation in the patient and output an EMG signal.
[0030] The control module acquires upper limb posture data through an inertial measurement unit and combines it with joint angle data. In passive training mode, it adjusts the power damping of the joint drive module based on the posture data and joint angle data. In active training mode, it adjusts the assist torque of the joint drive module in real time based on the posture data, joint angle data, and electromyographic signals.
[0031] In this embodiment, the control module includes multiple training modes, specifically passive charging mode, damping training mode, semi-active assist mode and active assist mode, which can be selected and switched through a human-computer interaction interface.
[0032] like Figure 4 and Figure 5 As shown, the control module is based on a dual-agent adaptive framework, specifically: the patient is regarded as a human intelligent agent, and the upper limb exoskeleton rehabilitation robot is regarded as a machine intelligent agent. Through multi-model control and reinforcement learning algorithms, the control parameters and training modes are updated online according to motion deviation, muscle load and comfort indicators.
[0033] Specifically: like Figure 4 and Figure 5 As shown, the control module can be divided into the following interrelated functional modules: The cognitive and strategic module of the human brain represents the patient's central nervous system, which is responsible for generating motor intentions and high-level strategies based on rehabilitation tasks and information such as vision, vestibular system, and proprioception, such as "raising your hand to a certain height" or "doing the movements according to the doctor's rhythm".
[0034] Neuromuscular System Module: This module represents the neural transmission and muscle fiber contraction process from the brain to the muscles. Motor commands issued by the brain are translated into muscle activation and the generation of torque within this module; this process can be partially observed using electromyography (EMG) sensors.
[0035] Human-machine coupled dynamics system: This refers to the dynamic system composed of the patient's upper limbs, soft tissues, exoskeleton mechanical structure, and connecting straps. This system is simultaneously subjected to physiological torques from the neuromuscular system and mechanical torques from the exoskeleton's motor / damping drive, serving as the physical carrier for the interaction between the "human intelligent agent" and the "machine intelligent agent."
[0036] Motor / damping drive module: This refers to the drive actuator installed at each joint of the exoskeleton. It can output assist torque in active mode, or provide adjustable damping and energy recovery in passive mode.
[0037] The exoskeleton controller module, comprising the aforementioned multi-model control and reinforcement learning algorithms, is the core of the "machine agent." It receives sensor information such as posture, joint angles, electromyographic signals, and interactive forces, and uses a dual-agent adaptive framework to calculate control parameters (assist torque, damping, training mode, etc.) to drive the motor / damping module.
[0038] Subjective evaluation / reward module: This module represents the patient's subjective evaluation of pain, fatigue, comfort, etc. during the training process. It can be input into the system via buttons, voice, or scales, or it can be internalized as part of the joint reward function.
[0039] The joint reward function module integrates motion deviation indicators, muscle load indicators, comfort indicators, and subjective evaluations into a reward signal for reinforcement learning. The joint reward function reflects both the control performance sought by the machine agent and the subjective feelings of the human agent, serving as an "evaluation benchmark" for the collaborative optimization of both.
[0040] The above modules form a closed loop through signal flow: human brain cognition and strategy generate movement intentions, which drive the upper limbs through the neuromuscular system and interact with the exoskeleton through human-machine coupled dynamics; sensors feed back posture, electromyography and interaction force information to the exoskeleton controller; the controller updates control parameters under the drive of multi-model reinforcement learning and joint reward function, and then acts on the exoskeleton through motor / damping drive to realize the mutual adaptive adjustment of human-machine dual intelligent agents.
[0041] This embodiment employs a dual-agent adaptive framework: the patient is considered Agent 1 (Human), and the upper limb exoskeleton is considered Agent 2 (Robot). The two interact physically through a human-machine coupled dynamics system, and adapt to each other under environmental / task constraints through multi-model control and reinforcement learning algorithms.
[0042] For Agent 1: Human, the patient obtains state information through vision, proprioception, etc. In the human brain's cognition and strategy module (Human Policy) The intention to move is formed in the brain, and muscle torque is generated through the neuromuscular system (Actuator). As a human-side movement This is applied to the human-machine coupled dynamic system. The patient simultaneously provides subjective evaluations / feedback based on subjective comfort, fatigue, pain, etc. Strategies for changing oneself through long-term learning and adaptability .
[0043] For Agent 2: Robot, the exoskeleton obtains its status through sensors such as joint angles, angular velocities, torques, and electromyography. Execute machine strategies in the exoskeleton controller The output control law generates mechanical torque via the motor / damping drive module (Actuator). Adjustable damping may be used as a mechanism side action. This applies to the same human-machine coupled dynamics system. Based on training task completion, physiological safety, and energy consumption, a joint reward function is constructed for the system. Used to update machine policies The parameters enable "machine-side adaptation: parameter update / ".
[0044] Human-machine coupled dynamics systems can be simplified as follows: ;in, The joint angle vector. , For velocity and acceleration, , , These are the equivalent inertia, damping, and stiffness matrices, respectively. This is the external disturbance torque.
[0045] On the Robot side, the control module of the upper limb exoskeleton visual rehabilitation robot has a pre-set set of training sub-models. = { , , …, Different sub-models Corresponding to different training modes and parameter combinations, such as "high-assistance, low-damping," "medium-assistance," and "low-assistance, high-damping (partially passive training)," each model contains a set of control parameters, mainly including: Joint-level parameters: (1) Assist torque ratio coefficient : Decision The magnitude of the auxiliary torque provided by the exoskeleton; (2) Impedance control parameters: stiffness Damping Adjusting the correction strength and interaction compliance; (3) damping coefficient Adjust the intensity of variable damping / generator braking in passive training mode.
[0046] Timing and mode parameters: (1) Intermittent training duty cycle: on-site duration Offset duration (2) Training mode weights or selection probability Used in different models (3) Difficulty level parameters: target trajectory amplitude, speed coefficient, number of repetitions, etc.
[0047] Safety and comfort related parameters: (1) Safety thresholds such as heart rate / muscle load / interaction force limit; (2) Weight coefficients that balance “exercise accuracy – muscle load – comfort” in the reward function.
[0048] The core of multi-model + reinforcement learning is to use feedback metrics during training to fine-tune these parameters and update model weights online, thereby realizing machine policies. Continuous optimization.
[0049] In this embodiment, motion deviation index : ;in, For the first Each joint at any time angular error, For the first The maximum allowable error or range for each joint. The number of joints involved in the control.
[0050] Muscle load index : ;in, , This refers to the number of channels in the electromyography (EMG) sensor. For the first Each electromyographic channel at time The root mean square value, For the first The maximum voluntary contraction of a muscle group For the first Normalized activation of individual muscle groups.
[0051] Comfort Index : ;in, , These are the weighting coefficients. Rate it as subjective comfort. Objective discomfort; ;in, , These are the weighting coefficients. Let t be the norm of the human-computer interaction force (or torque). The maximum acceptable human-computer interaction force (safety threshold). The rate of change of joint acceleration, The maximum acceptable level of jerkiness.
[0052] Joint reward function : + ;in, , , These are the weighting coefficients.
[0053] Figure 4 On the left is Agent 1: Human. The patient obtains their state through vision / proprioception. In the human brain's cognition and strategy module (Human Policy) Motor decisions are formed during the process, and actions are output via the neuromuscular system. and muscle torque It acts on the human-machine coupled dynamic system; and provides subjective rewards based on pain, fatigue, and comfort. .
[0054] Figure 4 On the right is Agent 2: Robot. The exoskeleton obtains its status through sensors. Machine Policy, an exoskeleton controller The control law is calculated by combining multi-model control and reinforcement learning algorithms, and the output action is generated through the motor / damping drive module. and torque This applies to the same human-machine coupled dynamics system. The joint reward function module is based on... , , and subjective rating calculation Used for updating and related control parameters.
[0055] Figure 5 The top section, "Environment / Task Requirements," outlines the desired rehabilitation or training task (target trajectory, range of motion, target heart rate zone, etc.); the middle section, "Human-Machine Physical Interaction," represents the human-machine coupled dynamic system; the left section, "Human Agent (Patient)," and the right section, "Machine Agent (Exoskeleton)," correspond to... Figure 4 Agent 1 and Agent 2 in the text. The bottom sections "Human-side adaptation: Learning / Plasticity" and "Machine-side adaptation: Parameter updating / ..." This means that both the human brain-neuro-muscle system and the exoskeleton controller can be continuously adjusted according to performance indicators during training, achieving bilateral adaptation.
[0056] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. 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 rehabilitation robot for the upper limbs with charging and discharging functions, characterized in that, include: The system comprises a mechanical body module, multiple joint drive modules, an energy storage module, a control module, and a sensor module. The mechanical body module is fixedly connected to the patient's upper limb to be rehabilitated. The joint drive modules have dual operating modes: a motor and a generator. In generator mode, the module generates electrical energy with the movement of the patient's upper limb and charges the energy storage module. In motor mode, the energy storage module powers the movement of the patient's upper limb. The energy storage module is electrically connected to the joint drive module, and the control module is electrically connected to both the joint drive module and the energy storage module. The sensor module is used to collect motion information of the patient's upper limb to be rehabilitated and feed it back to the control module.
2. The upper limb exoskeleton rehabilitation robot with charging and discharging function as described in claim 1, characterized in that, The joint drive module is mounted on the mechanical body module, which includes at least a shoulder joint unit and an elbow joint unit, and each joint drive module is mounted on the shoulder joint unit and the elbow joint unit respectively.
3. The upper limb exoskeleton rehabilitation robot with charging and discharging function as described in claim 1, characterized in that, The joint drive module includes a reversible motor-generator actuator and a speed reduction transmission mechanism connected to the reversible motor-generator actuator. The reversible motor-generator actuator adopts a permanent magnet synchronous motor and can switch between motor working mode and generator working mode.
4. The upper limb exoskeleton rehabilitation robot with charging and discharging function as described in claim 1, characterized in that, The energy storage module includes a rechargeable battery and a supercapacitor, used to store electrical energy generated in generator operating mode and to supply power to the joint drive module in motor operating mode.
5. The upper limb exoskeleton rehabilitation robot with charging and discharging function as described in claim 1, characterized in that, Each joint drive module is connected to the energy storage module via a DC bus; the control module obtains the voltage of the DC bus and realizes unified scheduling of energy between multiple joints based on the voltage of the DC bus.
6. The upper limb exoskeleton rehabilitation robot with charging and discharging function as described in claim 1, characterized in that, The sensor module includes an inertial measurement unit mounted on the mechanical body module and corresponding to the patient's upper arm or forearm. The inertial measurement unit is used to measure upper limb posture data.
7. The upper limb exoskeleton rehabilitation robot with charging and discharging function as described in claim 6, characterized in that, The control module acquires upper limb posture data through an inertial measurement unit and combines it with joint angle data. In passive training mode, it adjusts the power damping of the joint drive module based on the posture data and joint angle data. In active training mode, it adjusts the assist torque of the joint drive module in real time based on the posture data, joint angle data, and electromyographic signals.
8. The upper limb exoskeleton rehabilitation robot with charging and discharging function as described in claim 1, characterized in that, The sensor module also includes an electromyography (EMG) sensor attached to the relevant muscle groups of the patient's upper limb. The EMG sensor is used to detect the degree of muscle activation and output EMG signals.
9. The upper limb exoskeleton rehabilitation robot with charging and discharging function as described in claim 1, characterized in that, The control module is based on a dual-agent adaptive framework, specifically: the patient is regarded as a human agent and the upper limb exoskeleton rehabilitation robot is regarded as a machine agent. Through multi-model control and reinforcement learning algorithms, the control parameters and training modes are updated online according to motion deviation, muscle load and comfort indicators.
10. The upper limb exoskeleton rehabilitation robot with charging and discharging function as described in claim 1, characterized in that, The control module includes multiple training modes, specifically passive charging mode, damping training mode, semi-active assist mode, and active assist mode, which can be selected and switched through a human-computer interaction interface.