Gait phase recognition and assist control method and system

Abstract

This invention discloses a gait phase recognition and assist control method applicable to exoskeletons, comprising the following steps: acquiring motion data of the exoskeleton wearer's lower limbs through a sensor unit, the motion data including at least angular velocity data of the wearer's feet and angle data of the wearer's joints; initializing a finite state machine model, the finite state machine model containing multiple detached gait phase states; determining the state transition conditions of the finite state machine model based on the angular velocity data of the feet and the angle data of the joints to determine the wearer's current gait phase state; selecting a corresponding assist curve from a preset assist curve library based on the current gait phase state; generating control commands based on the assist curve and sending them to the actuator to output assist. This invention also provides a gait phase recognition and assist control system.

Description

Technical Field

[0001] This invention relates to the field of exoskeleton technology, and in particular to a gait phase recognition and assist control method and system for exoskeletons. Background Technology

[0002] With the aging population and increasing demand for rehabilitation medicine, assistive devices such as exoskeleton robots and rehabilitation robots have shown great potential in assisting patients with lower limb dysfunction in gait training and daily walking. One of the core technologies of these devices is the accurate identification of the wearer's gait phase and the provision of timely and appropriate assistance. However, existing gait recognition and assistive control methods generally have some technical shortcomings.

[0003] For example, WIRobotics' gait control method for exoskeleton robots predicts and controls the current gait state by recording and analyzing historical gait data. However, this method requires a large amount of memory to store historical data and incurs a high computational load during real-time processing, making it particularly difficult to run efficiently on resource-constrained microcontroller platforms. Furthermore, its control logic can be complex and relatively poorly interpretable, which is detrimental to the approval and clinical application of medical devices. Summary of the Invention

[0004] In view of the shortcomings of the prior art, the technical problem to be solved by the present invention is to provide a gait phase recognition and assist control method and system that is highly efficient, highly accurate and easy to interpret.

[0005] A gait phase recognition and assist control method, applicable to exoskeletons, includes the following steps: Step S1: Collect motion data of the exoskeleton wearer's lower limbs through the sensor unit. The motion data includes at least the angular velocity data of the wearer's feet and the angle data of the wearer's joints. Step S2: Initialize the finite state machine model, which contains multiple discrete state phase states; Step S3: Based on the angular velocity data of the foot and the angle data of the joint, determine the state transition conditions of the finite state machine model to determine the wearer's current gait phase state; Step S4: Based on the current gait phase state, select the corresponding assist curve from the preset assist curve library; and Step S5: Generate control commands based on the assist curve and send them to the actuator to output assist; The finite state machine model includes five discrete gait phase states, namely: heel strike, foot flat, support phase, heel off the ground, and swing phase. These five states cycle in sequence to form a complete gait cycle.

[0006] The sensor unit includes an inertial measurement unit and a Hall sensor. The inertial measurement unit is installed on the back of the wearer's foot to collect angular velocity data of the foot. The Hall sensor is installed at the joint axis to detect the angle data of the joint.

[0007] The state transition conditions in step S3 include: When the foot angular velocity drops from the peak dorsiflexion to below -5 degrees / second and the angle of the joint increases to above 5 degrees, the foot transitions from the heel-to-spot state to the foot-flat state. When the foot angular velocity is between -5 degrees / second and 5 degrees / second, and the angle of the joint increases to more than 10 degrees, the foot transitions from the flat state to the support phase state. When the foot angular velocity is greater than 5 degrees / second and the angle of the joint increases to more than 20 degrees, the foot transitions from the support phase state to the heel-off state. When the foot angular velocity reaches the peak plantar flexion and the joint angle increases to more than 30 degrees, the foot transitions from the heel-off state to the swing phase state; and When the foot angular velocity drops from the peak plantar flexion to below -10 degrees / second and the angle of the joint decreases to below 5 degrees, the foot transitions from the swing phase state to the heel strike state.

[0008] The assist curve is a torque-time function, and the assist curves corresponding to each gait phase state are as follows: The heel strike state: the assist torque increases linearly from 0 Nm to the first torque, and continues for the first time; With the foot flat on the ground: the assist torque is maintained between the first torque and the second torque for a second duration; The support phase state is as follows: the assist torque increases linearly from the second torque to the third torque, and continues for a third time. The heel-off-ground state: the assist torque linearly decreases from the third torque to the first torque, lasting for a fourth time period; and The swing phase state: the assist torque remains at 0 Nm for the fifth time; The first torque, the second torque, and the third torque increase sequentially, and the first time, the third time, the second time, the fourth time, and the fifth time increase sequentially.

[0009] Step S3 further includes an anti-shake mechanism: the transition from the current gait phase state to the target gait phase state is only executed when the state transition condition continuously meets a preset number of sampling periods.

[0010] In step S4, the assist curve is dynamically adjusted based on the user's real-time physiological feedback or real-time gait parameters. The real-time physiological feedback includes electromyography signals, heart rate, and fatigue level, while the real-time gait parameters include walking speed and stride length.

[0011] On the other hand, the present invention also provides a gait phase recognition and assist control system, applicable to exoskeletons, comprising: The sensor unit is used to collect motion data of the wearer's lower limbs, and the motion data includes at least angular velocity data of the foot and angle data of the joint; Finite state machine model parameter unit, used to store a finite state machine model containing multiple discrete phase states; An angular velocity threshold storage unit is used to store a preset angular velocity threshold. Angle threshold storage unit is used to store preset angle thresholds; A gait phase state transition determination unit, connected to the sensor unit, the angular velocity threshold storage unit, and the angle threshold storage unit, is used to determine the state transition conditions of the finite state machine model based on the angular velocity data of the foot and the angle data of the joint, combined with the angular velocity threshold and the angle threshold, so as to determine the current gait phase state. The assist curve library unit is used to store several assist curves; An assist curve selection unit, connected to the gait phase state transition determination unit and the assist curve library unit, is used to select a corresponding assist curve from the assist curve library unit according to the current gait phase state; and An execution unit, connected to the assist curve selection unit, is used to generate control commands based on the assist curve and send them to the actuator to output assist.

[0012] The finite state machine model includes five discrete gait phase states, namely: heel strike, foot flat, support phase, heel off the ground, and swing phase. These five states cycle in sequence to form a complete gait cycle.

[0013] The state transition conditions include: When the foot angular velocity drops from the peak dorsiflexion to below -5 degrees / second and the angle of the joint increases to above 5 degrees, the foot transitions from the heel-to-spot state to the foot-flat state. When the foot angular velocity is between -5 degrees / second and 5 degrees / second, and the angle of the joint increases to more than 10 degrees, the foot transitions from the flat state to the support phase state. When the foot angular velocity is greater than 5 degrees / second and the angle of the joint increases to more than 20 degrees, the foot transitions from the support phase state to the heel-off state. When the foot angular velocity reaches the peak plantar flexion and the joint angle increases to more than 30 degrees, the transition occurs from the heel-off state to the swing phase state; and When the foot angular velocity drops from the peak plantar flexion to below -10 degrees / second and the angle of the joint decreases to below 5 degrees, the foot transitions from the swing phase state to the heel strike state.

[0014] The assist curve is a torque-time function, and the assist curves corresponding to each gait phase state are as follows: The heel strike state: the assist torque increases linearly from 0 Nm to the first torque, and continues for the first time; With the foot flat on the ground: the assist torque is maintained between the first torque and the second torque for a second duration; The support phase state is as follows: the assist torque increases linearly from the second torque to the third torque, and continues for a third time. The heel-off-ground state: the assist torque linearly decreases from the third torque to the first torque, lasting for a fourth time period; and The swing phase state: the assist torque remains at 0 Nm for the fifth time; The first torque, the second torque, and the third torque increase sequentially, and the first time, the third time, the second time, the fourth time, and the fifth time increase sequentially.

[0015] This invention uses both angular velocity and joint angle sensors for cross-verification, combined with an anti-shake mechanism, resulting in more accurate gait assessment and smoother, more seamless assistance. Furthermore, this invention employs only four sensors, allowing for high-speed processing with a standard chip, resulting in rapid assistance response. The gait is divided into five stages: a gentle push upon landing, a powerful push-off, and a lift-off, perfectly corresponding to normal walking, making it simple, clear, and highly practical. Attached Figure Description

[0016] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a flowchart illustrating a preferred embodiment of the gait phase recognition and assist control method described in this invention.

[0017] Figure 2 This is a schematic diagram of a preferred embodiment of a gait phase recognition and assist control system according to the present invention. Detailed Implementation

[0018] The invention will now be further described with reference to the accompanying drawings.

[0019] See Figure 1 The gait phase recognition and assist control method provided in this embodiment of the invention is used for exoskeletons and includes the following steps: Step S1: Collect motion data of the wearer's lower limbs using a sensor unit. In this embodiment, the data includes at least the angular velocity of the foot and the angle data of the joint. The sensor unit includes at least an inertial measurement unit and a Hall sensor. The inertial measurement unit is installed on the dorsum of the wearer's foot to collect the angular velocity data of the foot; the Hall sensor is installed at the joint axis to detect the angle data of the joint. Specifically, the sensor unit may include two inertial measurement units and two Hall sensors.

[0020] In this embodiment, the inertial measurement unit can be a Bosch BMI270 model, which integrates a three-axis accelerometer and a three-axis gyroscope. These are installed on the dorsum of the wearer's left and right feet (e.g., on the shoe upper or ankle) to collect angular velocity data of the left and right feet, with a sampling frequency set to 100Hz. The Hall sensor can be an AMS AS5048A magnetic encoder with 14-bit resolution, installed at the left and right ankle joint axes of the exoskeleton (e.g., ankle or knee joints) to accurately detect real-time angle data of the left and right ankle joints.

[0021] In other embodiments, the inertial measurement unit can be mounted on the front of the lower leg or the side of the thigh, and the angular velocity of the foot can be derived using a kinematic model. The Hall sensor can be replaced with an optical encoder or a magnetostrictive sensor to provide higher accuracy or more robust joint angle detection.

[0022] Step S2: FSM state definition and initialization.

[0023] The processor's memory pre-stores finite state machine (FSM) model parameters, angular velocity thresholds, angle thresholds, and an assist curve library. In this embodiment, the processor can be an STMicroelectronics STM32F407VGT6 microcontroller unit (MCU) with a main frequency of 168MHz, 512KB Flash, and 128KB RAM. The preset angular velocity and angle thresholds are calibrated and optimized based on a large amount of normal gait data and rehabilitation gait data.

[0024] To better distinguish different gait phases, this invention defines an FSM containing five discrete gait phase states to describe a complete gait cycle: Heel strike (HS): The start of the gait cycle, when the heel contacts the ground; Flat foot position (FF): The soles of the feet are fully in contact with the ground; Support Phase (MS): The body's center of gravity is above the supporting leg, supporting the body on one leg; Heel off the ground (HO): The heel is lifted off the ground while the toes remain in contact with the ground; Swing phase (SW): The entire foot lifts off the ground and swings forward.

[0025] The FSM is initialized to the default state (usually heel-ground state HS) when the system starts up. Then the processor reads the sensor data in each sampling cycle and proceeds to the next step to determine the state transition.

[0026] To provide more precise assistance, in other implementations, the state definition of the FSM can be added or adjusted as needed in specific rehabilitation scenarios (such as climbing stairs, descending stairs, and running). For example, a "lift-off" state or more detailed states such as "accelerated swing" and "decelerated swing" can be added to accommodate more complex gait patterns. The angular velocity threshold and angle threshold in the state transition conditions can also be adaptively adjusted according to the individual differences of different users (such as height, weight, and gait characteristics) or personalized through machine learning algorithms.

[0027] S3: Gait phase state transition determination.

[0028] The processor is connected to the sensor unit and is used to receive and process the motion data in step S1. Based on the foot angular velocity data collected by the inertial measurement unit and the joint angle data collected by the Hall sensor in step S1, the processor determines the state transition conditions of the FSM. The specific state transition determination rules are as follows: Heel strike to foot flat position: When the foot angular velocity drops from the peak dorsiflexion (e.g., -150 degrees / second) to below -5 degrees / second, and the ankle angle detected by the Hall sensor increases from near 0 degrees (e.g., -5 degrees to 5 degrees) to above 5 degrees.

[0029] From the flat foot position to the support phase: when the foot angular velocity approaches 0 degrees / second (e.g., from -5 degrees / second to 5 degrees / second), and the ankle angle detected by the Hall sensor increases to more than 10 degrees.

[0030] Support phase to heel-off phase: When the foot angular velocity increases from near 0 degrees / second toward plantar flexion (e.g., greater than 5 degrees / second), and the ankle angle detected by the Hall sensor increases to more than 20 degrees.

[0031] Heel-off state to swing phase state: when the foot angular velocity reaches the peak plantar flexion (e.g., 100 degrees / second) and the ankle angle detected by the Hall sensor increases to more than 30 degrees.

[0032] Swing phase to heel strike: When the foot angular velocity drops from the peak of plantar flexion and reverses to the dorsiflexion direction (e.g., less than -10 degrees / second), and the ankle angle detected by the Hall sensor decreases to less than 5 degrees.

[0033] To avoid misjudgments and frequent switching caused by sensor noise or instantaneous fluctuations, other embodiments of the present invention may introduce a state transition anti-shake mechanism. That is, a transition from the current gait phase state to the next target gait phase state is determined only when the aforementioned state transition conditions are continuously satisfied for a preset number of sampling periods. In this embodiment, the preset number of sampling periods is 3 sampling periods, i.e., 30ms (sampling frequency 100Hz). The processor proceeds to step S4 only after the state transition determination is completed.

[0034] Step S4: Select the appropriate assist curve based on the currently determined gait phase state.

[0035] Based on the current gait phase state determined in step S3, the processor selects the corresponding assist curve from a pre-set assist curve library. These assist curves are torque-time functions, pre-designed and stored in the processor's assist curve library according to gait biomechanical principles and rehabilitation needs. The assist curves corresponding to each state are as follows: When the heel strikes and the foot is laid flat: the assist torque increases linearly from 0 Nm to the first torque (e.g., 5 Nm) and continues for the first time (e.g., 100 ms) to help the foot land smoothly.

[0036] From the flat foot position to the support phase: the assist torque is maintained between the first torque and the second torque (e.g., 10 Nm), slowly increased, and continued for a second time (e.g., 200 ms) to provide stable support.

[0037] From the support phase to the heel-off phase: the assist torque increases linearly from the second torque to the third torque (e.g., 15 Nm), and continues for a third time (e.g., 150 ms) to assist in lifting the heel.

[0038] From heel off the ground to the swing phase: the assist torque decreases linearly from the third torque to the first torque, lasting for a fourth time (e.g., 250ms), to assist the foot to leave the ground and swing.

[0039] Swing phase to heel strike phase: The assist torque is maintained at 0 Nm or at a minimum value for a fifth time (e.g., 300 ms) to allow free swinging and prepare for the next landing.

[0040] In this embodiment, the first torque, second torque, third torque, first time, second time, third time, and fourth time can all be adjusted according to the user's preferences and physical health. Generally speaking, the first torque, second torque, and third torque increase sequentially, and the first time, third time, second time, fourth time, and fifth time increase sequentially.

[0041] In other implementations, the assist curve may not be a preset fixed curve, but may be dynamically adjusted based on the user's real-time physiological feedback (such as electromyography signals, heart rate, and fatigue level) or gait parameters (such as walking speed and stride length). For example, when user fatigue is detected, the assist torque can be appropriately increased; when the user's walking speed increases, the duration of the assist curve can be adjusted.

[0042] The processor selects the corresponding assist curve from the assist curve library based on the current gait phase state, generates control commands, and then proceeds to step S5.

[0043] Step S5: Send control commands to the actuator to output assistance.

[0044] In step S5, the exoskeleton receives the control commands generated by the processor in step S4 and provides assistance to the wearer's lower limbs. In this embodiment, the assistive device mainly includes a motor and a reducer. The motor can be a Maxon EC-max 30 brushless DC motor (BLDC) with a rated torque of 20 Nm. The reducer can be a planetary gear reducer with a reduction ratio of 1:50, used to amplify the motor's output torque to meet the assistance requirements. The motor receives PWM signals from the processor through its controller to precisely control the motor's output torque.

[0045] After completing one assist output, the system returns to step S1 in the next sampling cycle to re-acquire motion data and repeat the cycle from step S1 to step S5, thereby achieving continuous recognition of the wearer's gait and real-time assist control.

[0046] Please refer to Figure 2 As shown, the present invention also provides a gait phase recognition and assist control system for exoskeletons, the preferred embodiment of which includes a sensor unit, an FSM model parameter unit, an angular velocity threshold storage unit, an angle threshold storage unit, an assist curve library unit, a gait phase state transition determination unit, an assist curve selection unit, and an execution unit.

[0047] The sensor unit is used to collect motion data of the wearer's lower limbs, including at least the angular velocity of the foot and the angle data of the joint. Specifically, the sensor unit includes at least an inertial measurement unit and a Hall sensor. The inertial measurement unit is installed on the dorsum of the wearer's foot to collect the angular velocity data of the foot; the Hall sensor is installed at the joint axis to detect the angle data of the joint.

[0048] The FSM model parameter unit is used to store at least five gait phase states, specifically: heel strike state: the start of the gait cycle, with the heel in contact with the ground; foot flat state: the sole of the foot is fully in contact with the ground; support phase state: the body's center of gravity is above the supporting leg, with single-leg support; heel lift-off state: the heel lifts off the ground, while the toes remain in contact with the ground; swing phase state: the entire foot lifts off the ground and swings forward.

[0049] The angular velocity threshold storage unit is used to store the angular velocity thresholds of the left and right feet. The preset angular velocity thresholds can be calibrated and optimized based on a large amount of normal gait data and rehabilitation gait data.

[0050] The angle threshold storage unit is used to store the angle thresholds of the left and right ankle joints. The preset angle thresholds can also be calibrated and optimized based on a large amount of normal gait data and rehabilitation gait data.

[0051] The gait phase state transition determination unit receives the collected foot angular velocity data and joint angle data to determine the current FSM transition state. The specific FSM transition determination rules are as follows: From heel strike to foot flattening: when the foot angular velocity drops from the peak dorsiflexion (e.g., -150 degrees / second) to below -5 degrees / second, and the ankle angle detected by the Hall sensor increases from near 0 degrees (e.g., -5 degrees to 5 degrees) to above 5 degrees; From foot flattening to stance: when the foot angular velocity approaches 0 degrees / second (e.g., -5 degrees / second to 5 degrees / second), and the ankle angle detected by the Hall sensor increases to above 10 degrees; From stance to heel lift: when the foot angular velocity drops from near 0 degrees / second to below -5 degrees / second, and the ankle angle detected by the Hall sensor increases to above 10 degrees; From stance to heel lift: when the foot angular velocity drops from near 0 degrees / second to below -5 degrees / second, and the ankle angle detected by the Hall sensor increases from below 10 degrees / second. The ankle angle detected by the Hall sensor increases from the plantar flexion direction to the swing phase when the foot angular velocity begins to increase (e.g., greater than 5 degrees / second) and increases to more than 20 degrees; from the heel-off state to the swing phase state when the foot angular velocity reaches the peak plantar flexion (e.g., 100 degrees / second) and increases to more than 30 degrees; from the swing phase state to the heel-striking state when the foot angular velocity falls back from the peak plantar flexion and reverses to the dorsiflexion direction (e.g., less than -10 degrees / second) and decreases to less than 5 degrees.

[0052] The assist curve library unit is used to store several assist curves, which are torque-time functions and can be pre-designed according to gait biomechanical principles and rehabilitation needs. In this embodiment, the assist curves corresponding to each state are as follows: From heel strike to foot flattening state: the assist torque increases linearly from 0 Nm to the first torque (e.g., 5 Nm) for a first time (e.g., 100 ms) to assist the foot in landing smoothly; From foot flattening state to support phase state: the assist torque is maintained between the first torque and the second torque (e.g., 10 Nm), increases slowly for a second time (e.g., 200 ms) to provide stable support; From support phase state to heel lift-off state: the assist torque increases linearly from the second torque to the third torque (e.g., 15 Nm) for a third time (e.g., 150 ms) to assist the heel to lift off; From heel lift-off state to swing phase state: the assist torque decreases linearly from the third torque to the first torque for a fourth time (e.g., 250 ms) to assist the foot to lift off and swing; From swing phase state to heel strike state: the assist torque is maintained at 0 Nm or a minimum value for a fifth time (e.g., 300 ms) to allow free swinging and prepare for the next landing.

[0053] The assist curve selection unit is used to select the corresponding assist curve from the assist curve library unit according to the currently determined gait phase state. The execution unit is used to generate corresponding control commands according to the corresponding assist curve and send them to the actuator to output assist.

[0054] This invention utilizes two sensors to collect two movement speeds of the wearer. Angular velocity quickly captures the foot's rotational motion in the air, while joint angle visually indicates whether the foot is flat on the ground or on tiptoe. The two signals corroborate each other. For example, when determining whether the foot has landed, it checks both whether the angular velocity has dropped to a certain value and whether the ankle joint angle has increased to a certain value. Furthermore, for higher accuracy, this invention incorporates an "anti-shake" mechanism that performs several consecutive checks before actually switching states. This effectively filters out interference signals from uneven surfaces or muscle tremors, resulting in a more natural and smooth walking experience.

[0055] Meanwhile, the technical solution of this invention uses only four sensors (two for each foot), collecting fewer types of data and shortening the signal processing link. This reduces the computational requirements on the chip; a regular microcontroller can complete the entire process in real time at a speed of 100 times per second, with virtually no delay in assistance. More importantly, this invention divides the gait cycle into five phases: heel strike, foot flattening, support phase, heel lift, and swing phase. This perfectly corresponds to the changes in the sole of the foot that a normal person intuitively feels while walking. The amount of assistance provided in each phase is also segmented. Simply put, a gentle cushioning force is applied upon landing, a stronger push is applied during the push-off, and the force is gradually reduced during the lift-off to allow the leg to swing freely. During the judgment process, it is immediately clear which value corresponds to which state, and the parameters can be flexibly adjusted according to different individuals, making it highly practical.

[0056] The above description merely illustrates preferred embodiments of the present invention and is quite specific and detailed; however, it should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept, and these all fall within the scope of protection of the present invention. Therefore, the scope of protection of this invention should be determined by the appended claims.

Claims

1. A gait phase recognition and assist control method, suitable for an exoskeleton, characterized in that, Includes the following steps: Step S1: Collect motion data of the exoskeleton wearer's lower limbs through the sensor unit. The motion data includes at least the angular velocity data of the wearer's feet and the angle data of the wearer's joints. Step S2: Initialize the finite state machine model, which contains multiple discrete state phase states; Step S3: Based on the angular velocity data of the foot and the angle data of the joint, determine the state transition conditions of the finite state machine model to determine the wearer's current gait phase state; Step S4: Select the corresponding assist curve from the preset assist curve library according to the current gait phase state; and Step S5: Generate control commands based on the assist curve and send them to the actuator to output assist.

2. The gait phase recognition and assist control method according to claim 1, characterized in that, The finite state machine model contains five discrete gait phase states, which are: heel strike, foot flat, support phase, heel off the ground, and swing phase. These five states cycle in sequence to form a complete gait cycle.

3. The gait phase recognition and assist control method according to claim 2, characterized in that, The sensor unit includes an inertial measurement unit and a Hall sensor. The inertial measurement unit is installed on the back of the wearer's foot to collect angular velocity data of the foot. The Hall sensor is installed at the joint axis to detect the angle data of the joint.

4. The gait phase recognition and assist control method according to claim 2, characterized in that, The state transition conditions in step S3 include: When the foot angular velocity drops from the peak dorsiflexion to below -5 degrees / second and the angle of the joint increases to above 5 degrees, the foot transitions from the heel-to-spot state to the foot-flat state. When the foot angular velocity is between -5 degrees / second and 5 degrees / second, and the angle of the joint increases to more than 10 degrees, the foot transitions from the flat state to the support phase state. When the foot angular velocity is greater than 5 degrees / second and the angle of the joint increases to more than 20 degrees, the foot transitions from the support phase state to the heel-off state. When the foot angular velocity reaches the peak plantar flexion and the joint angle increases to more than 30 degrees, the foot transitions from the heel-off state to the swing phase state; and When the foot angular velocity drops from the peak plantar flexion to below -10 degrees / second and the angle of the joint decreases to below 5 degrees, the foot transitions from the swing phase state to the heel strike state.

5. The gait phase recognition and assist control method according to claim 2, characterized in that, The assist curve is a torque-time function, and the assist curves corresponding to each gait phase state are as follows: The heel strike state: the assist torque increases linearly from 0 Nm to the first torque, and continues for the first time; With the foot flat on the ground: the assist torque is maintained between the first torque and the second torque for a second duration; The support phase state is as follows: the assist torque increases linearly from the second torque to the third torque, and continues for a third time. The heel-off-ground state: the assist torque linearly decreases from the third torque to the first torque, lasting for a fourth time period; and The swing phase state: the assist torque remains at 0 Nm for the fifth time; The first torque, the second torque, and the third torque increase sequentially, and the first time, the third time, the second time, the fourth time, and the fifth time increase sequentially.

6. The gait phase recognition and assist control method according to claim 1, characterized in that, Step S3 also includes an anti-shake mechanism: the transition from the current gait phase state to the target gait phase state is only executed when the state transition condition continuously meets the preset number of sampling periods.

7. The gait phase recognition and assist control method according to claim 1, characterized in that, The assist curve in step S4 is dynamically adjusted based on the user's real-time physiological feedback or real-time gait parameters. The real-time physiological feedback includes electromyography signals, heart rate, and fatigue level, while the real-time gait parameters include walking speed and stride length.

8. A gait phase recognition and assist control system, applicable to exoskeletons, characterized in that, include: The sensor unit is used to collect motion data of the wearer's lower limbs, and the motion data includes at least angular velocity data of the foot and angle data of the joint; Finite state machine model parameter unit, used to store a finite state machine model containing multiple discrete phase states; An angular velocity threshold storage unit is used to store a preset angular velocity threshold. Angle threshold storage unit is used to store preset angle thresholds; A gait phase state transition determination unit, connected to the sensor unit, the angular velocity threshold storage unit, and the angle threshold storage unit, is used to determine the state transition conditions of the finite state machine model based on the angular velocity data of the foot and the angle data of the joint, combined with the angular velocity threshold and the angle threshold, so as to determine the current gait phase state. The assist curve library unit is used to store several assist curves; The assist curve selection unit is connected to the gait phase state transition determination unit and the assist curve library unit, and is used to select the corresponding assist curve from the assist curve library unit according to the current gait phase state. and An execution unit, connected to the assist curve selection unit, is used to generate control commands based on the assist curve and send them to the actuator to output assist.

9. The gait phase recognition and assist control system according to claim 8, characterized in that, The finite state machine model contains five discrete gait phase states, which are: heel strike, foot flat, support phase, heel off the ground, and swing phase. These five states cycle in sequence to form a complete gait cycle.

10. The gait phase recognition and assist control system according to claim 9, characterized in that, The state transition conditions include: When the foot angular velocity drops from the peak dorsiflexion to below -5 degrees / second and the angle of the joint increases to above 5 degrees, the foot transitions from the heel-to-spot state to the foot-flat state. When the foot angular velocity is between -5 degrees / second and 5 degrees / second, and the angle of the joint increases to more than 10 degrees, the foot transitions from the flat state to the support phase state. When the foot angular velocity is greater than 5 degrees / second and the angle of the joint increases to more than 20 degrees, the foot transitions from the support phase state to the heel-off state. When the foot angular velocity reaches the peak plantar flexion and the joint angle increases to more than 30 degrees, the transition occurs from the heel-off state to the swing phase state; and When the foot angular velocity drops from the peak plantar flexion to below -10 degrees / second and the angle of the joint decreases to below 5 degrees, the foot transitions from the swing phase state to the heel strike state.

11. The gait phase recognition and assist control system according to claim 9, characterized in that, The assist curve is a torque-time function, and the assist curves corresponding to each gait phase state are as follows: The heel strike state: the assist torque increases linearly from 0 Nm to the first torque, and continues for the first time; With the foot flat on the ground: the assist torque is maintained between the first torque and the second torque for a second duration; The support phase state is as follows: the assist torque increases linearly from the second torque to the third torque, and continues for a third time. The heel-off-ground state: the assist torque linearly decreases from the third torque to the first torque, lasting for a fourth time period; and The swing phase state: the assist torque remains at 0 Nm for the fifth time; The first torque, the second torque, and the third torque increase sequentially, and the first time, the third time, the second time, the fourth time, and the fifth time increase sequentially.

Images