An interactive force perception-based lower limb rehabilitation robot control system and method

Abstract

The application relates to the technical field of medical rehabilitation robot control, and provides a lower limb rehabilitation robot control system and method based on interactive force sensing, which comprises the following modules: an external device communication module for receiving training task information and control instructions; a control mode management module for determining a target training mode and performing switching control; an interactive sensing strategy module for collecting sensing data and identifying a motion intention or determining a gait phase switching condition; a gait parameter adjustment module for adjusting parameters of a preset gait generation function; a gait trajectory generation module for generating target gait trajectories of lower limb joints; and a motion control module for performing interpolation processing and speed planning on the target gait trajectories. The application is used in the control process of a lower limb rehabilitation robot, and solves the technical problem that the training effect cannot meet actual requirements due to the lack of human-computer interactive sensing capability in the prior art.

Description

Technical Field

[0001] This application relates to the field of assistive medical rehabilitation robot control technology, and in particular to a lower limb rehabilitation robot control system and method based on interactive force perception. Background Technology

[0002] Currently, most lower limb rehabilitation robots are wearable rigid exoskeletons that control joint motor movements to propel the user through walking training. Existing technologies typically employ passive training methods, where each joint motor executes movements along a fixed trajectory, failing to adjust stride length and speed according to the individual's needs. Furthermore, the lack of human-machine interaction between the user and the robot results in inadequate training effectiveness. In addition, most existing solutions rely on manual control for starting, stopping, and speed adjustment, which presents significant difficulties and inconveniences for users with limited hand and foot mobility. Therefore, existing technologies suffer from a lack of human-machine interaction capabilities, leading to training outcomes that fail to meet practical requirements. Summary of the Invention

[0003] This application provides a lower limb rehabilitation robot control system and method based on interactive force perception, which solves the technical problem that the lack of human-computer interaction perception capability in the prior art leads to the training effect failing to meet actual needs.

[0004] To achieve the above objectives, this application adopts the following technical solution: Firstly, a lower limb rehabilitation robot control system based on interactive force perception is provided, comprising: an external device communication module for receiving training task information and control commands from an external device; the training task information includes training mode, step length parameters, step speed parameters, training duration, and target body parameters; the control commands include start, stop, pause, continue, acceleration, and deceleration commands; a control mode management module for determining a target training mode based on the training task information and switching between different training modes; and an interactive perception strategy module for collecting interactive force transmission data during training using the target training mode. The system utilizes sensor data from sensors, joint torque sensors, and plantar pressure sensors to identify the target object's movement intention or determine gait phase switching conditions based on the sensor data; a gait parameter adjustment module adjusts the parameters of a preset gait generation function according to the stride length parameters, stride speed parameters, and target object body parameters to obtain an adjusted gait generation function; a gait trajectory generation module generates target gait trajectories for each joint of the lower limb based on the adjusted gait generation function and outputs the expected joint angles at corresponding times; and a motion control module performs interpolation processing and speed planning on the target gait trajectory to generate continuous control commands.

[0005] In conjunction with the first aspect above, in one possible implementation, the system further includes: a safety monitoring module, used to detect abnormal states based on the joint torque sensor data and a preset safety threshold, and trigger protection control when an abnormality is detected; and a drive and sensor communication module, used to send continuous control commands to the joint actuator to drive the movement of the lower limb joints, and collect data from each sensor and feed it back to the interactive perception strategy module and the safety monitoring module.

[0006] In conjunction with the first aspect above, in one possible implementation, the target training mode includes an active training mode and a passive training mode; when the target training mode is an active training mode, the left and right legs are controlled to alternate gait movements according to the gait phase switching signal output by the interactive perception strategy module; when the target training mode is a passive training mode, the lower limb joints are controlled to move according to a preset trajectory according to the target gait trajectory output by the gait trajectory generation module.

[0007] In conjunction with the first aspect above, in one possible implementation, controlling the alternating gait movement of the left and right legs according to the gait phase switching signal output by the interaction perception strategy module includes: S1, determining whether the leg currently in the swing phase meets a preset phase switching trigger condition based on the interaction force sensor data, wherein the trigger condition is that the corresponding interaction force is greater than a preset force threshold; S2, generating a gait phase switching signal when the phase switching trigger condition is met; S3, switching the leg currently in the support phase to the swing phase and switching the leg currently in the swing phase to the support phase according to the gait phase switching signal; and repeatedly executing steps S1 to S3 to perform alternating gait movement of the left and right legs.

[0008] In conjunction with the first aspect above, in one possible implementation, controlling the lower limb joints to move along a preset trajectory based on the target gait trajectory output by the gait trajectory generation module includes: performing periodic interpolation processing on the desired angles of each joint in the target gait trajectory at the current moment to generate continuous joint position commands; performing velocity planning based on the joint position commands to obtain corresponding joint velocity commands, wherein the velocity planning includes trapezoidal velocity planning or S-shaped velocity planning; and driving each joint of the lower limb to move along the target gait trajectory based on the joint position commands and joint velocity commands.

[0009] In conjunction with the first aspect above, in one possible implementation, the step of identifying the target object's movement intention or determining the gait phase switching condition based on the sensing data includes: fusing data from the interaction force sensor, joint torque sensor, and plantar pressure sensor to obtain comprehensive sensing features; identifying the target object's movement intention based on the comprehensive sensing features, wherein the movement intention includes an intention to start, a intention to stop, an intention to accelerate, or an intention to decelerate; or, determining the current gait state based on the comprehensive sensing features and determining whether the gait phase switching condition is met; and outputting a gait phase switching signal when the gait phase switching condition is met.

[0010] In conjunction with the first aspect described above, in one possible implementation, adjusting the parameters of a preset gait generation function based on the stride length parameter, stride speed parameter, and target object body parameters to obtain an adjusted gait generation function includes: normalizing the stride length parameter, stride speed parameter, and target object body parameters to obtain standardized input parameters; adjusting the amplitude coefficient and frequency parameter in the preset gait generation function based on the standardized input parameters, wherein the amplitude coefficient is used to characterize the joint movement amplitude, and the frequency parameter is used to characterize the gait period; scaling the gait generation function according to the target object body parameters to match the lower limb length characteristics of the target object; and outputting the adjusted gait generation function.

[0011] In conjunction with the first aspect above, in one possible implementation, generating the target gait trajectory of each joint of the lower limb based on the adjusted gait generation function and outputting the expected joint angle at the corresponding time includes: substituting the current running time into the adjusted gait generation function to calculate the angle value of each joint at the corresponding time; continuously calculating the joint angle value at each time according to a preset control cycle to form a gait trajectory in which the joint angle changes over time; and outputting the joint angle value corresponding to the current control cycle as the expected angle of each joint.

[0012] In conjunction with the first aspect above, in one possible implementation, the gait generation function is a multi-level Fourier trigonometric series, including a hip joint gait generation function, a knee joint gait generation function, and an ankle joint gait generation function; wherein, the hip joint gait generation function satisfies the following formula:

[0013] in, Let t be the hip joint angle, t be the running time, stride be the stride parameter, and K be a constant. These are the coefficients of the Fourier trigonometric series. denoted as first-order frequency, and r as a gait-related parameter; the knee joint gait generation function and the ankle joint gait generation function adopt the same form of Fourier trigonometric series.

[0014] Secondly, a control method for a lower limb rehabilitation robot based on interactive force perception is provided, comprising: acquiring training task information and control commands, wherein the training task information includes training mode, stride length parameters, stride speed parameters, training duration, and target object body parameters; determining a target training mode based on the training task information, and switching between different target training modes; during training using the target training mode, collecting sensing data from interactive force sensors, joint torque sensors, and plantar pressure sensors, and identifying the target object's movement intention or determining gait phase switching conditions based on the sensing data; adjusting the parameters of a preset gait generation function based on the stride length parameters, stride speed parameters, and target object body parameters to obtain an adjusted gait generation function; generating target gait trajectories for each joint of the lower limb based on the adjusted gait generation function, and outputting the expected joint angles at corresponding times; performing interpolation processing and velocity planning on the target gait trajectory to generate continuous control commands.

[0015] This application provides a lower limb rehabilitation robot control system and method based on interactive force perception. Through an interactive perception strategy module, it collects and fuses multi-source sensor data such as interactive force, joint torque, and plantar pressure. This enables accurate identification of the target object's movement intention or determination of gait phase switching conditions, thereby achieving gait switching control based on the target object's intention in active training mode. This solves the problem of existing technologies lacking human-computer interaction perception capabilities. Simultaneously, through a gait parameter adjustment module, the gait generation function is personalized according to the target object's body parameters. Combined with trajectory generation and motion control, this ensures the smoothness and adaptability of the training trajectory, improving rehabilitation training effects and the target object's experience. This addresses the technical problem in existing technologies where the lack of human-computer interaction perception capabilities leads to training effects that cannot meet actual needs.

[0016] It should be understood that the descriptions of technical features, technical solutions, beneficial effects, or similar language in this application do not imply that all features and advantages can be achieved in any single embodiment. Rather, it is understood that the description of a feature or beneficial effect means that a specific technical feature, technical solution, or beneficial effect is included in at least one embodiment. Therefore, the descriptions of technical features, technical solutions, or beneficial effects in this specification do not necessarily refer to the same embodiment. Furthermore, the technical features, technical solutions, and beneficial effects described in this embodiment can be combined in any suitable manner. Those skilled in the art will understand that embodiments can be implemented without one or more specific technical features, technical solutions, or beneficial effects of a particular embodiment. In other embodiments, additional technical features and beneficial effects may be identified in specific embodiments that do not embody all embodiments. Attached Figure Description

[0017] Figure 1 A system architecture diagram of a lower limb rehabilitation robot control system based on interactive force perception is provided for embodiments of this application; Figure 2 A flowchart illustrating a lower limb rehabilitation robot control system based on interactive force sensing, provided for an embodiment of this application; Figure 3 A flowchart illustrating another lower limb rehabilitation robot control system based on interactive force sensing provided in this application embodiment; Figure 4 A flowchart illustrating another lower limb rehabilitation robot control system based on interactive force sensing provided in this application embodiment; Figure 5 A flowchart illustrating another lower limb rehabilitation robot control system based on interactive force sensing provided in this application embodiment; Figure 6 This is a flowchart illustrating a lower limb rehabilitation robot control method based on interactive force perception, provided as an embodiment of this application. Detailed Implementation

[0018] In the description of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B. The "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone. Furthermore, "at least one" means one or more, and "multiple" means two or more. The terms "first," "second," etc., do not limit the quantity or order of execution, and "first," "second," etc., do not necessarily imply differences.

[0019] It should be noted that, in this application, the terms "exemplary" or "for example" are used to indicate that something is being described as an example, illustration, or illustration. Any embodiment or design described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or design solutions. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0020] The lower limb rehabilitation robot control method based on interactive force sensing provided in this application embodiment can be applied to, for example... Figure 1 In the lower limb rehabilitation robot control system shown, based on interactive force sensing, such as Figure 1As shown, the system includes: an external device communication module 101, a control mode management module 102, an interactive sensing strategy module 103, a gait parameter adjustment module 104, a gait trajectory generation module 105, and a motion control module 106. The modules interact with each other via a data bus or communication network to collaboratively complete lower limb rehabilitation training tasks.

[0021] The external device communication module 101 is used to receive training task information and control commands from external devices.

[0022] The training task information includes training mode, stride length parameters, stride speed parameters, training duration, and target body parameters. The control commands include start, stop, pause, continue, accelerate, and decelerate commands.

[0023] In one possible implementation, the external device communication module serves as the interface for information exchange between the system and the outside world. Its communication methods can include wired communication (such as USB or serial port) or wireless communication (such as WiFi or Bluetooth). The specific form of the external device is diverse. For example, it could be a human-computer interface terminal operated by medical personnel to input specific training parameters; or it could be a brain-computer interface device worn by the target individual. This device collects and analyzes the bioelectrical signals from the target individual's cerebral cortex, converts them into control commands such as start and stop, and sends them to the system, thus providing a convenient operating method for target individuals with upper limb mobility impairments. Through this module, the system can flexibly acquire multi-dimensional task requirements, laying the foundation for subsequent personalized control.

[0024] The control mode management module 102 is used to determine the target training mode based on the training task information and to switch between different training modes.

[0025] In one possible implementation, the target training mode includes active training and passive training modes. In passive training mode, the robot leads the movement, and the target object follows; in active training mode, the target object's subjective movement intention is reflected through interaction, and the robot assists in following. This module calls the corresponding control strategy interface based on the parsing results to ensure that the system's operating state is consistent with the preset task, thus achieving adaptive support for various rehabilitation scenarios.

[0026] The interactive perception strategy module 103 collects sensing data from the interactive force sensor, joint torque sensor, and plantar pressure sensor during the training process using the target training mode, and identifies the target object's movement intention or determines the gait phase switching condition based on the sensing data.

[0027] In one possible implementation, the interactive perception strategy module is key to achieving human-computer interaction. Interactive force sensors are typically positioned at the points where the robot's leg structure contacts the human leg to sense the active force applied by the target object; joint torque sensors are integrated into the joint drive unit to provide feedback on joint load; and plantar pressure sensors are located on the sole of the foot to determine the ground contact state. This module reads data from these multi-source sensors in real time and, through data fusion or threshold judgment logic, identifies the target object's intention to accelerate, decelerate, or take a step, thereby converting the target object's active awareness into control signals for the robot. This solves the problem of passive participation and poor interactive experience in traditional solutions.

[0028] The gait parameter adjustment module 104 is used to adjust the parameters of the preset gait generation function according to the stride length parameter, stride speed parameter and target object body parameters to obtain the adjusted gait generation function.

[0029] In one possible implementation, a fixed gait trajectory cannot meet the needs of all target subjects due to differences in height, leg length, and rehabilitation stage. The gait parameter adjustment module receives personalized parameters from external input and modifies the pre-stored general gait generation function within the system. For example, it adjusts the stride coefficient based on the target subject's leg length parameter and the gait speed parameter based on the rehabilitation progress. Through this parameterized adjustment mechanism, the system can generate a gait baseline adapted to the target subject's body characteristics, ensuring training comfort and safety.

[0030] The gait trajectory generation module 105 is used to generate target gait trajectories for each joint of the lower limb based on the adjusted gait generation function, and output the expected joint angles at the corresponding times.

[0031] In one possible implementation, the gait trajectory generation module, based on an adjusted function model and incorporating a time variable, calculates in real-time the expected angular positions (i.e., the positions of key joints such as the hip and knee joints) that should be reached at the current moment. The trajectory data output by this module is a discrete sequence of angle points, providing precise positional targets for motion control and ensuring that the robot's movements conform to human physiological principles.

[0032] The motion control module 106 is used to perform interpolation processing and speed planning on the target gait trajectory and generate continuous control commands.

[0033] In one possible implementation, since the desired angles output by the gait trajectory generation module are typically discrete points, directly sending them to the actuator may cause motion jitter. The motion control module fills in intermediate values ​​between the discrete points using an interpolation algorithm and performs velocity planning (such as limiting acceleration), transforming the discrete angle points into smooth and continuous position and velocity commands. This process ensures the smoothness of the robot's movements and avoids secondary damage to the target object caused by mechanical impacts.

[0034] This application embodiment constructs a complete closed-loop control system. From receiving and parsing external commands to recognizing intent from multi-source sensor data, and then to generating and smoothing personalized gait, all modules work collaboratively to achieve adaptive rehabilitation training based on the interactive force perception of the target object. This not only solves the problem of the lack of human-computer interaction perception capabilities in existing technologies, but also meets the personalized needs of the target object through parameterized adjustments, effectively improving the effect and experience of rehabilitation training.

[0035] In one possible implementation, the lower limb rehabilitation robot control system based on interactive force perception provided in this embodiment further includes a safety monitoring module and a drive and sensor communication module.

[0036] The safety monitoring module is used to detect abnormal states based on joint torque sensor data and preset safety thresholds, and to trigger protection control when an abnormality is detected.

[0037] In one possible implementation, during rehabilitation training, the target individual may experience abnormal joint stress due to muscle spasms, fatigue, or improper operation. The safety monitoring module reads joint torque sensor data at each joint in real time and compares it to preset safety thresholds. These preset safety thresholds are not fixed values ​​but can be personalized based on the target individual's rehabilitation stage and physical condition. For example, the system can assign different torque limits based on the spasticity level (high, medium, low) set in the training task. When joint torque data exceeds the corresponding safety threshold, the system determines it as an abnormal state (such as leg spasms), and the safety monitoring module immediately triggers protective controls. Specific forms of protective control may include, but are not limited to: immediately stopping the current training task (emergency stop), switching to a flexible control mode to mitigate impact, or issuing an audible and visual alarm to prompt medical personnel to intervene. This torque feedback-based safety mechanism effectively prevents secondary injuries and ensures the absolute safety of the target individual during training.

[0038] The drive and sensor communication module is used to send continuous control commands to the joint actuator to drive the movement of the lower limb joints, and to collect data from various sensors and feed it back to the interactive sensing strategy module and the safety monitoring module.

[0039] In one possible implementation, this module acts as a bridge between the upper-level control strategy and the lower-level hardware devices. To ensure the real-time nature of control commands and the synchronization of sensor data, the drive and sensor communication modules typically employ high-real-time industrial bus protocols. In the command transmission path, continuous control commands (such as position and speed commands) generated by the motion control module are packaged into data frames conforming to the communication protocol and sent to each joint actuator to drive the motors to perform corresponding rotational movements, thereby actuating the lower limb joints. In the data transmission path, this module is responsible for collecting raw data from various sensor nodes, including interactive force sensors, joint torque sensors, plantar pressure sensors, and motor encoders, and then parsing and feeding it back to the upper layer. This feedback data is used by the interactive perception strategy module for intent recognition and gait phase determination, and by the safety monitoring module for abnormal state detection. Through this bidirectional closed-loop communication link of command transmission and data transmission, the system achieves precise perception and control of the robot's motion state.

[0040] As an example, during the robot's "stand up" action, the drive and sensor communication module continuously collects data from the plantar pressure sensor. When the plantar pressure value reaches a preset ground contact threshold, the system determines that the robot is in an upright position and stops the standing action. At the same time, the safety monitoring module monitors the torque values ​​of the knee and hip joints throughout the process. If a sudden change in torque is detected during the standing process, an emergency stop protection is immediately triggered.

[0041] In one possible implementation of the embodiments of this application, combined with Figure 1 The target training modes in the aforementioned control mode management module 102 include active training mode and passive training mode.

[0042] like Figure 2 As shown, when the target training mode is active training mode, the alternating movement of the left and right legs is controlled by the gait phase switching signal output by the interactive perception strategy module. This can be achieved through the following S1 to S3, which are explained in detail below: S1. Based on the data from the interactive force sensor, determine whether the leg currently in the swing phase meets the preset phase switching trigger condition.

[0043] Interactive force sensors are typically placed at the points where the robot's leg structure contacts the human leg, such as the inner sides of the thigh and calf levers, to sense the active force applied by the target object. The swing phase refers to the stage where the leg leaves the ground and swings forward in the air. If the target object intends to take a step at this time, it will actively pull its leg forward, thereby generating a continuous forward interactive force on the sensor. The trigger condition is that the corresponding interactive force is greater than a preset force threshold.

[0044] In one possible implementation, the system collects real-time interaction force data of the swinging leg and compares its absolute value or component in a specific direction with a preset force threshold. This preset force threshold is not fixed and can be personalized according to the target individual's muscle strength level and rehabilitation stage, for example, set to a value within the range of 5 to 20 Newtons. Only when the interaction force continuously exceeds this threshold for a certain period of time (e.g., 200 milliseconds) is it considered a valid trigger, in order to avoid false triggers caused by shaking.

[0045] It should be noted that the judgment process can also be aided by data from joint torque sensors or plantar pressure sensors. For example, simultaneously detecting whether the plantar pressure of the supporting leg is stable can improve the accuracy of the judgment.

[0046] As an example, when the target's left leg is in the support phase and its right leg is in the swing phase, the system continuously monitors the readings of the right leg's interaction force sensor. When the target wants to step out with its right leg, it will subconsciously exert force to lift its right leg. Once this interaction force exceeds a preset force threshold, the system determines that the right leg meets the phase switching trigger condition.

[0047] Based on the above steps, the system can keenly capture the subtle movement intentions of the target object, realizing the transformation from machine controlling human to human driving machine, which greatly enhances the target object's enthusiasm for participating in training.

[0048] S2. When the phase switching trigger condition is met, a gait phase switching signal is generated.

[0049] Among them, the gait phase switching signal is a state transition instruction inside the system, used to trigger the state machine to flip.

[0050] In one possible implementation, after confirming that the triggering condition is met, the interactive sensing strategy module generates a level transition signal or a specific data frame identifier and sends it to the control mode management module and the motion control module. This signal marks the end of the current gait cycle and the beginning of the next cycle.

[0051] It should be noted that, in order to ensure the smoothness of the switching, the system can also predict the motion trend at the next moment based on the current magnitude and direction of the interaction force while generating the switching signal, so as to provide feedforward compensation for subsequent motion control.

[0052] Based on the above steps, the system transforms the biomechanical intent of the target object into machine-recognizable control commands, thus establishing a bridge for human-computer interaction.

[0053] S3. Based on the gait phase switching signal, switch the leg currently in the support phase to the swing phase, and switch the leg currently in the swing phase to the support phase.

[0054] The support phase refers to the stage where the legs contact the ground and support the body weight. The phase switching between the two legs is alternating, that is, the two states of left support / right swing and right support / left swing alternately.

[0055] In one possible implementation, upon receiving the switching signal, the motion control module immediately adjusts the control strategy. The leg in the original swing phase (e.g., the right leg) stops swinging and switches to a support control strategy, controlling the motor to output high torque to support the body weight; the leg in the original support phase (e.g., the left leg) releases the support lock and switches to a swing control strategy, coordinating with the target's lifting action to take a step forward. At this point, the system enters the next gait cycle. Steps S1 to S3 are repeated in this manner, alternating between left and right leg gait movements to achieve target-driven walking training.

[0056] It should be noted that at the moment of switching, the system will adjust the joint stiffness through an impedance control algorithm to ensure the stability of the center of gravity transfer process and prevent the target object from falling due to instability.

[0057] The system of this application embodiment realizes gait training that is completely driven by the target object's intention. It can not only exercise the target object's limb motor ability, but also promote the functional reorganization of the nervous system through active participation, thus effectively improving the effect of rehabilitation training.

[0058] like Figure 3 As shown, when the target training mode is passive training mode, the lower limb joints are controlled to move along a preset trajectory based on the target gait trajectory output by the gait trajectory generation module. This can be achieved through the following steps S301 to S303, which are explained in detail below: S301. Perform periodic interpolation on the desired angles of each joint in the target gait trajectory at the current moment to generate continuous joint position commands.

[0059] Periodic interpolation refers to the process of inserting an intermediate angle value between two discrete desired angle points according to a preset control period (such as 1 millisecond to 10 milliseconds).

[0060] In one possible implementation, since the desired angles output by the gait trajectory generation module are typically discrete key points calculated based on the gait cycle, directly sending them to the driver would result in a stepped motion. The motion control module uses linear interpolation or spline interpolation algorithms to calculate smooth transition points between adjacent desired angles, thereby generating a continuous sequence of joint position commands. For example, during the knee joint's movement from 0 degrees to 90 degrees, the interpolation algorithm subdivides it into thousands of tiny position increments to ensure continuous motor rotation.

[0061] It should be noted that the accuracy of interpolation directly affects the smoothness of motion. The shorter the control cycle, the smoother the generated trajectory, but the higher the computational performance requirements of the controller.

[0062] Based on the above steps, abrupt changes between discrete points on the trajectory are eliminated, providing a basic position command flow for smooth joint movement.

[0063] S302. Based on the joint position command, perform velocity planning to obtain the corresponding joint velocity command.

[0064] Velocity planning refers to designing the velocity curve during joint movement to control acceleration and jerk, thereby avoiding mechanical shock. In the embodiments of this application, velocity planning includes trapezoidal velocity planning or S-shaped velocity planning.

[0065] In one possible implementation, position commands alone are insufficient; planning how the robot reaches the target position is also necessary. Trapezoidal velocity planning refers to a velocity curve that is trapezoidal, including three stages: uniform acceleration, uniform speed, and uniform deceleration. It is computationally simple but suffers from sudden acceleration changes, potentially causing slight vibrations. S-shaped velocity planning, on the other hand, refers to a velocity curve that is S-shaped, with continuously changing acceleration, resulting in smoother motion, but requiring slightly higher computational power. The system can automatically select the planning method based on the target object's rehabilitation level; for example, S-shaped velocity planning is prioritized for highly sensitive target objects.

[0066] It should be noted that speed planning is not only applied to single-step motion, but also to the start-stop process, to ensure that the robot accelerates slowly when starting and decelerates slowly when stopping, to prevent the target object from falling due to inertia.

[0067] Based on the above steps, the refined design of the velocity curve effectively suppresses mechanical vibration and impact during motion, thereby improving the comfort of the target object.

[0068] S303: Based on joint position commands and joint velocity commands, drive each joint of the lower limb to move according to the target gait trajectory.

[0069] The joint position command and the joint speed command together constitute the control input of the actuator, realizing dual closed-loop control of position and speed.

[0070] In one possible implementation, the motion control module packages the generated position and velocity commands and sends them to each joint actuator via the drive and sensor communication module. The position and velocity loop controllers within the actuators precisely adjust the motor current according to the commands, driving the joint motors to execute movements accurately. This coordinated control of position and velocity ensures both the accuracy of trajectory tracking and limits fluctuations in movement speed, thus guaranteeing the safety and stability of passive training.

[0071] It should be noted that during the movement, the system will still monitor the joint torque sensor data in real time. Once abnormal resistance is detected, the speed plan will be adjusted or the protection will be triggered immediately to achieve flexible control in passive mode.

[0072] Based on the above steps, the transformation from discrete trajectories to continuous smooth motion in passive training mode was realized, ensuring the quality and safety of rehabilitation training.

[0073] This application's embodiments combine periodic interpolation processing with trapezoidal or S-shaped velocity planning to solve the motion jitter problem caused by direct control of discrete trajectories, ensuring the stability and continuity of robot motion in passive training mode, and providing a safe and comfortable passive rehabilitation experience for the target object.

[0074] In one possible implementation of the embodiments of this application, combined with Figure 1 ,like Figure 4 As shown, in the above-mentioned interactive perception strategy module 103, the identification of the target object's motion intention or the determination of gait phase switching conditions based on sensor data can be specifically implemented through the following S401 and S402, which are explained in detail below: S401. The data from the interaction force sensor, joint torque sensor, and plantar pressure sensor are fused to obtain comprehensive sensing features.

[0075] Among them, the interaction force sensor data reflects the changes in contact force between the target object's limbs and the robot structure, the joint torque sensor data reflects the load state of each joint, and the plantar pressure sensor data reflects the contact between the foot and the ground. The comprehensive perception feature refers to the numerical or vector representation of the target object's motion state extracted after processing the above multi-source heterogeneous data.

[0076] In one possible implementation, the data fusion process is not a simple data stacking, but rather uses specific algorithmic models to eliminate measurement errors or noise interference that may exist in a single sensor. The system first preprocesses the acquired raw data, including filtering, noise reduction, and normalization operations to eliminate dimensional differences. The system can employ a feature-level fusion strategy to construct a multi-dimensional feature vector from the interaction force components, joint torque values, and plantar pressure values ​​at the same time. To improve recognition accuracy, the system can also assign different weights to data from different sensors. The weight coefficients can be adaptively adjusted based on training patterns or historical data of the target object, thereby obtaining more accurate comprehensive perception features.

[0077] It should be noted that, compared to relying solely on a single interactive force sensor for judgment, multi-sensor fusion can effectively solve the problem of misjudgment caused by target object shaking, sensor loosening, or environmental interference. For example, when the interactive force sensor detects an abnormal force signal, if the plantar pressure sensor shows that the foot is not touching the ground and the joint torque change is stable, the system can determine that the force signal is a interference signal rather than a true movement intention, thereby improving the robustness of the system.

[0078] S402. Based on comprehensive perception features, identify the movement intention of the target object. Or, based on comprehensive perception features, determine the current gait state and whether the gait phase switching condition is met; when the gait phase switching condition is met, output the gait phase switching signal.

[0079] Motion intention includes initiation intention, stopping intention, acceleration intention, or deceleration intention. Recognizing motion intention is the process of mapping comprehensive sensory features into specific control commands. Gait state determination, on the other hand, is based on features to determine whether the current phase is the support phase or the swing phase, and then decides whether to switch between them.

[0080] In one possible implementation, the specific logic for intent recognition and phase determination can be dynamically configured according to different training modes. In one implementation, the system can establish an intent recognition model for recognizing intentions such as starting, stopping, accelerating, and decelerating. For example, when the "forward-backward interaction force" component in the comprehensive perception features continuously increases while the "plantar pressure" component remains stable, the system recognizes it as an acceleration intention; when the "forward-backward interaction force" component increases in the opposite direction, it recognizes it as a deceleration intention. In another implementation, the system employs a multi-condition constraint mechanism for determining gait phase switching conditions. For example, the condition for determining the transition from the swing phase to the support phase no longer depends solely on the interaction force threshold, but requires simultaneously satisfying two conditions: "the interaction force is greater than a preset force threshold" and "the plantar pressure sensor detects a ground contact signal." This multi-condition constraint mechanism can effectively prevent abnormal phase switching caused by accidental triggering of interaction forces when the target object's legs swing in the air.

[0081] It should be noted that intent recognition models can be rule-based decision logic or classification models based on machine learning (such as support vector machines (SVM) or neural networks).

[0082] As an example, during the target object's active walking training, the system calculates comprehensive sensory features in real time. When it detects that the interaction force feature value of the target object's right leg exceeds a threshold, and the plantar pressure feature value of the right foot decreases (indicating foot lift), while the right knee joint torque feature value conforms to the swing trend, the system determines that the gait phase switching condition is met, outputs a switching signal, and controls the right leg to switch from the support phase to the swing phase. This process fully demonstrates the advantages of multi-source data fusion in complex motion recognition.

[0083] This application embodiment utilizes multi-sensor fusion technology to not only accurately identify the movement intention of the target object, but also ensures the reliability of gait phase switching through a multi-condition constraint mechanism. This effectively solves the problems of one-sided and easily interfered information perceived by a single sensor, further improving the safety and smoothness of rehabilitation training.

[0084] In one possible implementation of the embodiments of this application, combined with Figure 1 ,like Figure 5 As shown, the gait parameter adjustment module 104 described above can be implemented through the following steps S501 to S504, which are explained in detail below: S501. Normalize the stride length parameter, stride speed parameter, and target object body parameter to obtain standardized input parameters.

[0085] Among them, stride length is usually measured in meters, stride speed is measured in meters per second, and body parameters include height, leg length, etc. The units are different, and directly inputting them into the model will lead to inconsistent calculation units.

[0086] In one possible implementation, the system first acquires raw parameter values ​​input by medical personnel or collected by sensors, such as the target object's leg length being 0.8 meters and the target stride length being 0.5 meters. These parameters are then divided by their corresponding standard reference values; for example, leg length is divided by the average leg length of a standard adult (e.g., 0.85 meters), and stride length is divided by a standard stride length range related to height. This maps all input parameters to the interval [0,1] or [-1,1], resulting in dimensionless standardized input parameters. This processing method eliminates numerical fluctuations caused by differences in the body shape of different target objects, making the subsequent parameter adjustment algorithm more universal.

[0087] Based on the above steps, the differences between different physical dimensions were eliminated, laying a data foundation for subsequent parameter adjustments based on a unified model.

[0088] S502. Based on standardized input parameters, adjust the amplitude coefficient and frequency parameters in the preset gait generation function.

[0089] The preset gait generation function contains several undetermined coefficients, which determine the shape of the gait trajectory. The amplitude coefficient directly affects the angle of joint swing, while the frequency parameter determines the time required to complete one gait cycle. The amplitude coefficient is used to characterize the joint movement amplitude, and the frequency parameter is used to characterize the gait cycle.

[0090] In one possible implementation, the system pre-stores a baseline gait generation function fitted from a large amount of normal human gait data. When standardized parameters are input, the system adjusts the amplitude coefficients and frequency parameters in the function using linear mapping or a lookup table. For example, if the target object sets a large stride length parameter, the system will correspondingly increase the amplitude coefficients of hip extension and flexion to increase stride length; if the target object sets a large gait speed parameter, the system will increase the frequency parameter and shorten the gait cycle, thereby increasing walking speed.

[0091] It should be noted that the parameter adjustment process is not a simple linear scaling, but rather requires consideration of joint kinematic constraints. The system can preset multiple sets of adjustment coefficient mapping tables, and look up the optimal amplitude coefficients and frequency parameters according to different combinations of standardized parameters to ensure that the adjusted gait trajectory meets both the needs of the target object and the physiological laws of movement.

[0092] Based on the above steps, personalized customization of gait trajectories is achieved, enabling rehabilitation training programs to accurately match the training intensity requirements of different target individuals.

[0093] S503. Based on the target object's body parameters, scale the gait generation function to match the target object's lower limb length characteristics.

[0094] Among them, scale correction refers to scaling the gait trajectory geometrically according to the actual anatomical parameters of the target object, such as leg length, thigh length, and calf length, in order to solve the problem of differences in the end trajectory of target objects of different heights.

[0095] In one possible implementation, after adjusting the amplitude and frequency, the system further introduces a scale correction factor. This factor is typically the ratio of the target object's leg length to the standard model's leg length. The system multiplies this scale correction factor into the angle output of the gait generation function or the trajectory calculation formula, ensuring that the generated joint angles or Cartesian coordinate trajectories are adapted to the target object's limb length. For example, for a target object with shorter legs, at the same joint angles, its foot movement trajectory will be shorter; this difference can be compensated for through scale correction.

[0096] Preferably, the gait generation function is a multi-level Fourier trigonometric series, including a hip joint gait generation function, a knee joint gait generation function, and an ankle joint gait generation function; The hip joint gait generation function satisfies the following formula:

[0097] in, Let t be the hip joint angle, t be the running time, stride be the stride parameter, and K be a constant. These are the coefficients of the Fourier trigonometric series. Here, r represents the first-order frequency, and r is a gait-related parameter. The knee joint gait generation function and the ankle joint gait generation function adopt the same form of Fourier trigonometric series. The specific parameters of the hip joint gait generation function, the knee joint gait generation function, and the ankle joint gait generation function can be set according to actual needs, and this application embodiment does not limit this.

[0098] It should be noted that in this formula, stride, as a global scaling factor, directly determines the stride size; K is the baseline offset, corresponding to the hip joint angle when the human body is standing. The product of r and r determines the fundamental frequency of gait, i.e. the walking speed; These are the amplitude coefficients of each harmonic level, which together determine the shape characteristics of the gait curve, such as the degree of extension in the mid-stance phase and the peak flexion at the end of the swing phase. The function forms of the knee and ankle joints are similar, only the coefficients are different. In this application embodiment, a 7-level Fourier series is preferably used, which is the best balance between computational efficiency and fitting accuracy: too low a series (such as 3 levels) will result in an overly smooth trajectory curve, losing the detailed features of human walking (such as the slight impact when the heel touches the ground); too high a series (such as 10 levels or more) will significantly increase the computational burden of the controller and easily introduce high-frequency noise. Through the above algorithm, the gait trajectory generation module substitutes the current running time into the adjusted gait generation function to calculate the angle values ​​of each joint at the corresponding time; according to the preset control cycle, it continuously calculates the joint angle values ​​at each time, forming a gait trajectory in which the joint angles change over time; and outputs the joint angle value corresponding to the current control cycle as the expected angle of each joint. This real-time generation method based on analytical equations has higher flexibility than the lookup table method, can respond to parameter adjustment commands in real time, and realize the dynamic and smooth transition of the gait trajectory.

[0099] Based on the above steps, the matching degree between gait trajectory and target body characteristics is further improved, avoiding training discomfort or joint damage risks caused by limb length mismatch.

[0100] S504. Output the adjusted gait generation function.

[0101] The adjusted gait generation function contains all the parameter information adapted to the current target object's body characteristics and training task requirements, and is a deterministic mathematical model.

[0102] In one possible implementation, the system substitutes the parameter values, after amplitude adjustment, frequency adjustment, and scale correction, into a preset function structure to generate the final executable gait generation function. This function uses time t as the independent variable and joint angle as the dependent variable, and can output the desired angle at any time in real time.

[0103] It should be noted that the output gait generation function can be a set of analytical expressions or a set of discrete parameter data packets, which can be called by the gait trajectory generation module.

[0104] Based on the above steps, the transformation from abstract parameters to a specific mathematical model was completed, providing a precise calculation basis for subsequent trajectory generation.

[0105] Furthermore, in one possible implementation, the current running time is substituted into the adjusted gait generation function to calculate the angle values ​​of each joint at the corresponding moment; according to the preset control cycle, the joint angle values ​​at each moment are continuously calculated to form a gait trajectory in which the joint angles change over time; and the joint angle value corresponding to the current control cycle is output as the desired angle of each joint.

[0106] To address the technical problem that existing technologies lack human-computer interaction perception capabilities, resulting in training effects that fail to meet practical needs, this application provides a lower limb rehabilitation robot control method based on interactive force perception. Figure 6 A flowchart illustrating the lower limb rehabilitation robot control method based on interactive force sensing provided in this application embodiment is shown below. Figure 6 As shown, the method includes: S601, Obtain training task information and control commands.

[0107] The training task information includes training mode, stride length parameters, stride speed parameters, training duration, and target body parameters.

[0108] In one possible implementation, the system needs to complete initialization and parameter configuration before rehabilitation training begins. Medical staff or the target individual inputs the training prescription through an external device (such as a human-computer interface or brain-computer interface device). The training task information includes not only basic motion parameters (such as a stride length of 0.5 meters and a walking speed of 20 steps / minute) but also the target individual's body parameters (such as a height of 170cm and a leg length of 75cm). These parameters are the foundational data for generating personalized gait trajectories. Control commands are intervention instructions received in real time during training, such as acceleration intentions sent by the target individual through the brain-computer interface or pause commands sent manually via a button. The system listens for and receives this data in real time through a communication interface, storing it in a memory buffer for subsequent modules to access.

[0109] S602. Determine the target training mode based on the training task information, and switch between different target training modes.

[0110] In one possible implementation, the system parses the task information and extracts the mode field. If the field indicates an active training mode, the system configures a control strategy of intent-triggered, i.e., waiting for the target object to exert force to trigger gait switching; if the field indicates a passive training mode, the system configures a control strategy of trajectory following, i.e., the robot completely dominates the movement. The mode switching control logic ensures that the system can flexibly adapt to the needs of the target object at different training stages, for example, using passive mode during the warm-up stage and switching to active mode during the active participation stage. It should be understood that mode switching usually occurs between training tasks to avoid the risk of sudden changes during movement.

[0111] S603. During the training process using the target training mode, the sensor data of the interactive force sensor, joint torque sensor and plantar pressure sensor are collected, and the movement intention of the target object or the gait phase switching condition is identified based on the sensor data.

[0112] In one possible implementation, during training, the system reads data from various sensors at high frequency via a low-level driver bus (such as EtherCAT). For active training mode, the system focuses on monitoring interaction force sensor data to determine if the interaction force of the current swinging leg exceeds a preset threshold, thus determining whether the gait phase switching condition is met. For passive training mode, the system can fuse joint torque and plantar pressure data to identify whether the target object intends to accelerate or decelerate, thereby fine-tuning the motion parameters. This process transforms the mechanical signals of the physical world into digital features that can be processed by control logic, achieving real-time capture of the target object's motion intentions.

[0113] S604. Based on the stride length parameter, stride speed parameter, and target object body parameter, adjust the parameters of the preset gait generation function to obtain the adjusted gait generation function.

[0114] In one possible implementation, to ensure that the generated gait trajectory conforms to human kinematics and is adapted to the target object's body shape, the system needs to modify the built-in mathematical model. This step first normalizes the heterogeneous input parameters to eliminate dimensional differences; then, it adjusts the amplitude coefficient of the function based on the stride length parameter and the frequency parameter based on the stride speed parameter; finally, it scales the function output based on the target object's leg length parameter. Through this series of adjustments, the preset general gait function is transformed into a personalized function specific to the current target object, ensuring the trajectory's rationality and comfort from the outset.

[0115] S605. Based on the adjusted gait generation function, generate the target gait trajectory of each joint of the lower limb and output the expected joint angle at the corresponding time.

[0116] In one possible implementation, within each control cycle, the system substitutes the current running time into the adjusted gait generation function. The function uses trigonometric series operations to calculate the desired angle values ​​of the hip, knee, and ankle joints in real time. These angle values ​​constitute discrete target gait trajectory points. The system arranges these trajectory points in chronological order to form a continuous sequence of joint angles, which serves as the baseline command for motion control and is output to the next stage.

[0117] S606: Perform interpolation processing and velocity planning on the target gait trajectory to generate continuous control commands.

[0118] In one possible implementation, since the desired angle obtained by S605 is discrete in time, direct output may cause motor jitter. This step uses an interpolation algorithm (such as linear interpolation or spline interpolation) to fill in intermediate points between two desired angle points to ensure the continuity of position commands. Simultaneously, to prevent mechanical shock during start-up and shutdown, the system uses trapezoidal or S-shaped velocity planning to smooth the motion velocity curve and limits acceleration and jerk. Finally, the smoothed position and velocity commands are packaged into continuous control commands and sent to the joint actuator to drive the robot to smoothly move the target object's lower limbs.

[0119] This application's embodiments construct a complete methodological closed loop from task input, intent perception, trajectory planning to underlying execution. This method not only achieves intelligent control based on interactive force perception, but also ensures the personalization and safety of rehabilitation training through parameter adjustment and trajectory planning algorithms, thereby improving the clinical applicability and training effectiveness of the rehabilitation robot.

[0120] This application has been described herein with reference to various embodiments. However, in implementing the claimed application, those skilled in the art can understand and implement other variations of the disclosed embodiments by reviewing the accompanying drawings, the disclosure, and the appended claims. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude multiple components. A single processor or other unit can implement several functions listed in the claims. Although different dependent claims may recite certain measures, this does not mean that these measures cannot be combined to produce good results.

[0121] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and drawings are merely exemplary illustrations of this application as defined by the appended claims, and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from the spirit and scope of this application. Thus, if such modifications and modifications of this application fall within the scope of the claims of this application and their equivalents, this application is also intended to include such modifications and modifications.

Claims

1. A lower limb rehabilitation robot control system based on interactive force perception, characterized in that, include: The external device communication module is used to receive training task information and control commands from external devices; The training task information includes training mode, step length parameters, step speed parameters, training duration, and target body parameters; the control commands include start, stop, pause, continue, accelerate, and decelerate commands. The control mode management module is used to determine the target training mode based on the training task information and to switch between different training modes. The interactive perception strategy module collects sensor data from interactive force sensors, joint torque sensors, and plantar pressure sensors during training using the target training mode, and identifies the target object's movement intention or determines gait phase switching conditions based on the sensor data. The gait parameter adjustment module is used to adjust the parameters of the preset gait generation function according to the stride length parameter, stride speed parameter and target object body parameters, so as to obtain the adjusted gait generation function; The gait trajectory generation module is used to generate target gait trajectories for each joint of the lower limb based on the adjusted gait generation function, and output the expected joint angles at the corresponding times. The motion control module is used to perform interpolation processing and velocity planning on the target gait trajectory and generate continuous control commands.

2. The system according to claim 1, characterized in that, The system also includes: The safety monitoring module is used to detect abnormal states based on the joint torque sensor data and preset safety thresholds, and to trigger protection control when an abnormality is detected. The drive and sensor communication module is used to send continuous control commands to the joint actuator to drive the movement of the lower limb joints, and to collect data from various sensors and feed it back to the interactive sensing strategy module and the safety monitoring module.

3. The system according to claim 1, characterized in that, The target training mode includes active training mode and passive training mode; When the target training mode is active training mode, the left and right legs are controlled to alternate gait based on the gait phase switching signal output by the interactive perception strategy module. When the target training mode is passive training mode, the lower limb joints are controlled to move according to the preset trajectory based on the target gait trajectory output by the gait trajectory generation module.

4. The system according to claim 3, characterized in that, The step of controlling the alternating movement of the left and right legs based on the gait phase switching signal output by the interactive perception strategy module includes: S1. Based on the data from the interactive force sensor, determine whether the leg currently in the swing phase meets the preset phase switching trigger condition, wherein the trigger condition is that the corresponding interactive force is greater than the preset force threshold. S2. When the phase switching trigger condition is met, a gait phase switching signal is generated; S3. According to the gait phase switching signal, switch the leg currently in the support phase to the swing phase, and switch the leg currently in the swing phase to the support phase; Repeat steps S1 to S3 above, alternating between left and right leg gait movements.

5. The system according to claim 3, characterized in that, The step of controlling the lower limb joints to move along a preset trajectory based on the target gait trajectory output by the gait trajectory generation module includes: Periodic interpolation is performed on the desired angles of each joint in the target gait trajectory at the current moment to generate continuous joint position commands. Based on the joint position command, velocity planning is performed to obtain the corresponding joint velocity command, wherein the velocity planning includes trapezoidal velocity planning or S-shaped velocity planning; Based on the joint position and joint velocity commands, the joints of the lower limbs are driven to move according to the target gait trajectory.

6. The system according to claim 1, characterized in that, The process of identifying the target object's movement intention or determining gait phase switching conditions based on the sensor data includes: The data from the interaction force sensor, joint torque sensor, and plantar pressure sensor are fused to obtain comprehensive sensing features. Based on the comprehensive perception features, the motion intention of the target object is identified, wherein the motion intention includes an intention to start, an intention to stop, an intention to accelerate, or an intention to decelerate; or, Based on the comprehensive sensing features, the current gait state is determined, and it is determined whether the gait phase switching conditions are met; When the gait phase switching condition is met, a gait phase switching signal is output.

7. The system according to claim 1, characterized in that, The step of adjusting the parameters of the preset gait generation function based on the stride length parameter, stride speed parameter, and target object body parameters to obtain the adjusted gait generation function includes: The stride length parameter, stride speed parameter, and target body parameter are normalized to obtain standardized input parameters. Based on the standardized input parameters, the amplitude coefficient and frequency parameter in the preset gait generation function are adjusted, wherein the amplitude coefficient is used to characterize the joint movement amplitude, and the frequency parameter is used to characterize the gait period. Based on the target object's body parameters, the gait generation function is scaled to match the target object's lower limb length characteristics; Output the adjusted gait generation function.

8. The system according to claim 1, characterized in that, The process of generating target gait trajectories for each joint of the lower limb based on the adjusted gait generation function, and outputting the expected joint angles at corresponding times, includes: Substitute the current running time into the adjusted gait generation function to calculate the angle values ​​of each joint at the corresponding moment; According to the preset control cycle, the joint angle values ​​at each moment are continuously calculated to form the gait trajectory of joint angle changes over time. The joint angle value corresponding to the current control cycle is used as the expected angle output for each joint.

9. The system according to claim 7, characterized in that, The gait generation function is a multi-level Fourier trigonometric series, including a hip joint gait generation function, a knee joint gait generation function, and an ankle joint gait generation function; The hip joint gait generation function satisfies the following formula: in, Let t be the hip joint angle, t be the running time, stride be the stride parameter, and K be a constant. These are the coefficients of the Fourier trigonometric series. denoted as first-order frequency, and r as a gait-related parameter; the knee joint gait generation function and the ankle joint gait generation function adopt the same form of Fourier trigonometric series.

10. A control method for a lower limb rehabilitation robot based on interactive force perception, characterized in that, include: Acquire training task information and control commands, wherein the training task information includes training mode, step length parameter, step speed parameter, training duration and target object body parameters; The target training mode is determined based on the training task information, and switching control is performed between different target training modes; During the training process using the target training mode, sensor data from interactive force sensors, joint torque sensors, and plantar pressure sensors are collected, and the movement intention of the target object or the gait phase switching conditions are identified based on the sensor data. Based on the stride length parameter, stride speed parameter, and target object body parameter, the parameters of the preset gait generation function are adjusted to obtain the adjusted gait generation function; Based on the adjusted gait generation function, the target gait trajectory of each joint of the lower limb is generated, and the expected joint angle at the corresponding time is output. The target gait trajectory is interpolated and velocity is planned to generate continuous control commands.

Images