Somatosensory driving control method and system of scooter

Abstract

The invention relates to the technical field of driving control, and discloses a somatosensory driving control method and system of a scooter. The method comprises the following steps: acquiring an original somatosensory signal of the scooter; performing resonance amplification and multi-mode frequency mixing on the original somatosensory signal to obtain a somatosensory characteristic signal; establishing a scooter-rider coupling system dynamic model based on the somatosensory characteristic signal, and performing real-time state calculation to obtain a current state parameter and a state deviation value; performing motion sensing body weight distribution change mapping conversion on the current state parameter and the state deviation value to obtain a riding control demand parameter; and calculating a real-time driving control signal sequence according to the riding control demand parameters, and carrying out front and rear wheel motor torque dynamic distribution on the real-time driving control signal sequence to obtain a target motor driving execution instruction set. The problem of delay in traditional somatosensory control is eliminated, and safe and stable motor driving control is realized by combining dynamic torque distribution and a safety monitoring mechanism of road adhesion conditions.

Description

Technical Field

[0001] The present invention relates to the technical field of drive control, and in particular to a somatosensory drive control method and system for a scooter. Background Art

[0002] Although body-sensing controlled scooters are currently beginning to adopt weight distribution sensing technology, most systems use simple pressure sensors and basic linear mapping algorithms, which have problems such as large signal noise, significant delayed response, and low control accuracy. These systems usually use post-processing data analysis methods, filtering and interpreting the data through software algorithms after data collection. This not only introduces additional time delays, but also makes it difficult to capture subtle body sensory inputs. At the same time, existing technologies mostly use integer-order controllers with fixed parameters, which cannot be adaptively adjusted for different riding scenarios and rider habits, resulting in limited understanding of riding intentions by the control system, affecting the overall riding experience. In addition, traditional body-sensing control systems fail to fully consider the coupled dynamic characteristics of the scooter and the rider, and simply regard the two as independent systems, resulting in unstable control and even safety risks when riding at high speeds or in complex road conditions. Summary of the Invention

[0003] The main purpose of the present invention is to provide a somatosensory drive control method and system for a scooter. The present invention eliminates the delay problem in traditional somatosensory control and combines dynamic torque distribution and safety monitoring mechanisms based on road adhesion conditions to achieve safe and stable motor drive control.

[0004] To achieve the above object, the present invention provides a somatosensory drive control method for a scooter, comprising the following steps: Collect the original somatosensory signals of the scooter; Performing resonance amplification and multi-modal mixing on the original somatosensory signal to obtain a somatosensory characteristic signal; Establishing a scooter-rider coupling system dynamics model based on the somatosensory characteristic signal, and performing real-time state calculation to obtain current state parameters and state deviation values; Performing a somatosensory weight distribution change mapping conversion on the current state parameter and the state deviation value to obtain a riding control demand parameter; A real-time drive control signal sequence is calculated according to the riding control demand parameter, and the front and rear wheel motor torques are dynamically distributed to the real-time drive control signal sequence to obtain a target motor drive execution instruction set.

[0005] Optionally, in a first implementation of the first aspect of the present invention, collecting the original somatosensory signal of the scooter includes: A micro-accelerometer array is installed on the pedal surface of the scooter, pressure sensors are set on both sides of the pedal, and a six-axis inertial measurement unit is assembled at the central control unit to obtain a somatosensory data acquisition network; The body sensing data acquisition network is used to acquire the rider's weight distribution change data and the scooter's posture data in real time to obtain the original sensing signal.

[0006] Optionally, in a second implementation of the first aspect of the present invention, performing resonant amplification and multimodal mixing on the original somatosensory signal to obtain a somatosensory characteristic signal includes: Inputting the original body sensory signal into an LC resonant circuit array, and adjusting the inductance and capacitance parameters of the LC resonant circuit array so that the resonant frequency matches the characteristic frequency range of human motion, thereby obtaining a resonantly amplified body sensory signal; Performing a combined operation on the resonantly amplified body sensory signals through a multimodal physical mixing circuit to obtain a comprehensive characteristic signal of weight distribution changes; Inputting the comprehensive characteristic signal of weight distribution change into a phase-locked amplifier circuit, calculating a phase ratio between a reference signal in the phase-locked amplifier circuit and the comprehensive characteristic signal of weight distribution change, and amplifying a valid signal in the comprehensive characteristic signal of weight distribution change according to the phase ratio, thereby obtaining a body sensing signal with a high signal-to-noise ratio; The direction and acceleration information of the weight distribution change are extracted according to the high signal-to-noise ratio somatosensory signal to obtain a somatosensory characteristic signal.

[0007] Optionally, in a third implementation of the first aspect of the present invention, establishing a scooter-rider coupling system dynamics model based on the somatosensory characteristic signal and performing real-time state calculation to obtain current state parameters and state deviation values includes: Extracting the position coordinates, center of gravity distribution, and posture data of the rider on the pedal based on the somatosensory feature signals, and combining them with the scooter body parameters to obtain a state observation dataset; Based on the state observation data set, a system configuration description matrix of the scooter and the rider is constructed and a scooter-rider coupled system dynamics model including a mass matrix, a Coriolis force matrix and a gravity term is derived; Constructing a mirror state estimator based on the scooter-rider coupling system dynamics model, inputting the somatosensory characteristic signal and inertial measurement unit data into the mirror state estimator, performing state prediction and measurement update using an extended Kalman filter algorithm to obtain current state parameters; A state deviation value between the ideal state and the current state is calculated based on the current state parameters and a preset ideal riding model.

[0008] Optionally, in a fourth implementation of the first aspect of the present invention, constructing a system configuration description matrix of the scooter and rider based on the state observation dataset and deriving a scooter-rider coupled system dynamics model including a mass matrix, a Coriolis force matrix, and a gravity term includes: Combining the center of mass position, orientation angle, and rider posture angle of the scooter into a system configuration description matrix based on the state observation data set; Calculating mass attributes of the system configuration description matrix, analyzing mass distribution and inertia parameters of various parts of the scooter and the rider, and obtaining a system mass matrix; Calculating velocity coupling characteristics based on the system configuration description matrix and the system mass matrix, and constructing a Coriolis force matrix through operations on velocity derivatives and angular velocity; A gravitational potential energy analysis is performed on the system configuration description matrix to calculate the gravity term, and the system mass matrix, the Coriolis force matrix and the gravity term are combined to construct a scooter-rider coupling system dynamics model.

[0009] Optionally, in a fifth implementation of the first aspect of the present invention, a mirror state estimator is constructed according to the dynamic model of the scooter-rider coupling system, the somatosensory characteristic signal and inertial measurement unit data are input into the mirror state estimator, and state prediction and measurement update are performed through an extended Kalman filter algorithm to obtain current state parameters, including: A mirror state estimator is constructed based on the dynamic model of the scooter-rider coupling system, the system state variables are organized into a state vector containing position, angle and its derivatives, and a state transfer function and a measurement function are set to obtain a state space description; Performing time update processing on the state vector in the state space description, and calculating the state prediction value and the prediction covariance matrix according to the system dynamics equation to obtain preliminary predicted state data; Using the somatosensory characteristic signal and the inertial measurement unit data as observation inputs, calculating the deviation between the preliminary predicted state data and the actual measurement value, and calculating the Kalman gain coefficient based on the measurement noise matrix and the prediction covariance matrix to obtain a state correction factor; The state correction factor is applied to correct the preliminary predicted state data, and the system state covariance matrix is updated to obtain the current state parameters.

[0010] Optionally, in a sixth implementation of the first aspect of the present invention, performing a somatosensory weight distribution change mapping conversion on the current state parameter and the state deviation value to obtain a riding control demand parameter includes: Analyzing a change pattern of the rider's weight distribution according to the current state parameters, determining a center of gravity offset and a change rate in the front, rear, left, and right directions based on the change pattern, and constructing a weight distribution change vector according to the center of gravity offset and the change rate; Performing error function calculation on the weight distribution change vector and the state deviation value to obtain a control deviation matrix; Deconstructing the control deviation matrix and calculating the control law, and converting the control deviation matrix into the required driving force and steering force to obtain basic driving demand data; According to the basic driving demand data and the current operating state of the scooter, the Jacobian matrix is calculated and combined with the proportional-differential controller to generate the riding control demand parameters.

[0011] Optionally, in a seventh implementation of the first aspect of the present invention, calculating a real-time drive control signal sequence according to the riding control requirement parameters, and dynamically allocating front and rear wheel motor torques on the real-time drive control signal sequence to obtain a target motor drive execution instruction set includes: Inputting the riding control demand parameter into a fractional-order elastic controller, and calculating a fractional-order control signal by setting order parameters of a fractional-order differential operator and a fractional-order integral operator in the fractional-order elastic controller and combining proportional, integral, and differential gain coefficients; Establishing an elastic response control instruction between speed control and position control based on the fractional-order control signal; Performing Lyapunov energy function constraints and time delay correction on the elastic response control instructions to obtain a real-time drive control signal sequence; According to the road adhesion condition and current state of the scooter, the front and rear wheel torques are distributed to the real-time drive control signal sequence to obtain a target motor drive execution instruction set.

[0012] Optionally, in an eighth implementation of the first aspect of the present invention, the front and rear wheel torques are distributed on the real-time drive control signal sequence according to the road adhesion condition and current state of the scooter to obtain a target motor drive execution instruction set, including: Analyzing the adhesion characteristics of the current road surface on which the scooter is operating to obtain an adhesion characteristic index, and calculating the front and rear wheel torque distribution coefficient based on the adhesion characteristic index combined with the scooter's wheel speed and acceleration data; Decomposing the real-time drive control signal sequence based on the front and rear wheel torque distribution coefficients, combining the longitudinal control component and the lateral control component according to the set steering coefficient, and respectively calculating the front wheel torque command and the rear wheel torque command; Applying a torque change rate limit to the front wheel torque command and the rear wheel torque command to obtain a torque output value with a smooth transition; The smoothly transitioned torque output value is compared with the current safety state, and the safe control mode is smoothly switched to under abnormal conditions to obtain a target motor drive execution instruction set.

[0013] The present invention also provides a somatosensory drive control system for a scooter, comprising: The acquisition module is used to collect the original body sensory signals of the scooter; an amplification module, configured to perform resonance amplification and multi-modal mixing on the original somatosensory signal to obtain a somatosensory characteristic signal; A state calculation module is used to establish a scooter-rider coupling system dynamics model based on the somatosensory characteristic signal and perform real-time state calculation to obtain current state parameters and state deviation values; a mapping conversion module, configured to perform a somatosensory weight distribution change mapping conversion on the current state parameter and the state deviation value to obtain a riding control demand parameter; The dynamic allocation module is used to calculate a real-time drive control signal sequence according to the riding control requirement parameters, and dynamically allocate the front and rear wheel motor torques of the real-time drive control signal sequence to obtain a target motor drive execution instruction set.

[0014] In summary, the technical solution provided by the present invention constructs a micro-accelerometer array with a triangular lattice topology on the pedal surface, combined with pressure sensors on both sides of the pedal and a six-axis inertial measurement unit in the center, forming a multi-dimensional, high-density sensor network that accurately captures changes in the rider's weight distribution and eliminates the blind spots existing in traditional linear array arrangements. Using an LC resonant circuit array and multimodal physical mixing technology, somatosensory signals are processed at the circuit physics level. By matching the resonant frequency with the characteristic frequency of human motion and using phase-locked amplification technology, the signal-to-noise ratio is significantly improved, effectively suppressing environmental noise interference. A dynamic model of the scooter-rider coupling system is established, encompassing the mass matrix, Coriolis force matrix, and gravity terms. Using a mirror state estimator and an extended Kalman filter algorithm, accurate estimation of the system state is achieved, avoiding the direct interpretation of user intent in traditional methods. Using weight distribution change mapping technology, the rider's center of gravity movement is naturally converted into speed and direction control requirements. By deconstructing the control law calculation, an intuitive conversion from somatosensory input to control output is achieved, allowing the rider to obtain a natural control experience without learning. By setting the parameters of the fractional-order differential and integral operators, a fractional-order elastic controller is employed to achieve smooth response characteristics intermediate between those of traditional controllers. Combined with elastic drive mapping, this allows the scooter to adaptively respond to the rider's sensory input. Lyapunov energy function constraints and delay-compensated feedforward technology ensure that system control outputs remain within a safe range and eliminate the latency issues associated with traditional sensory control. Furthermore, dynamic torque distribution tailored to road adhesion conditions and a safety monitoring mechanism ensure safe and stable motor drive control. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 1 is a schematic diagram of the steps of a somatosensory drive control method for a scooter according to an embodiment of the present invention; Figure 2 This is a structural block diagram of a somatosensory drive control system for a scooter in one embodiment of the present invention.

[0016] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0018] Reference Figure 1 This embodiment provides a somatosensory drive control method for a scooter, comprising the following steps: S1, collects the original somatosensory signals of the scooter; A somatosensory data acquisition network is constructed, using the scooter's pedals as the physical carrier. This network integrates multiple sensor components and achieves system-level fusion through a unified signal conditioning and acquisition path. An array of 12 miniature accelerometers is evenly distributed across the scooter's pedal surface. This array adopts a triangular lattice topology, with 5 cm spacing between each accelerometer. This ensures that any weight distribution area is covered by multiple sensing points, effectively avoiding false positives caused by blind spots in a single sensor. Each accelerometer has a high sensitivity of 0.001g and a high sampling rate of 1000Hz. The measurement range is set to ±2g, ensuring accurate response to subtle pressure changes on the pedal surface and achieving high temporal resolution data updates. To enhance the ability to capture lateral pressure distribution characteristics, a high-precision pressure sensor is placed on each side of the pedal. These pressure sensors have a minimum detectable force of 0.05N and support a linear output range of 0-1000N. This provides real-time information on the load changes between the rider's left and right feet, providing data support for control logic such as riding tendency determination and steering intention recognition. At the same time, a six-axis inertial measurement unit (IMU) is installed in the scooter's central control unit. This unit integrates a three-axis accelerometer and a three-axis gyroscope to dynamically obtain the scooter's attitude parameters, including changes in roll, pitch, and yaw angles. The IMU complements the acceleration array and jointly captures the scooter's overall attitude and local force changes. All sensors are connected to the central signal conditioning circuit via shielded differential cables. This circuit performs primary filtering and gain amplification on all analog sensor signals, standardizing them into a uniform millivolt voltage output, which is then input into the central processing module through an analog-to-digital conversion channel. Through the somatosensory data acquisition network, multimodal raw data including the rider's foot pressure distribution, acceleration response, and posture changes are acquired in real time to form raw sensor signals reflecting the rider's behavioral intentions and the scooter's dynamic response.

[0019] S2, resonant amplification and multi-modal mixing of the original somatosensory signal to obtain a somatosensory characteristic signal; Specifically, the raw sensory signal is input into an LC resonant circuit array, which consists of multiple parallel resonant branches. Each branch uses a tuned inductor and capacitor to create a resonant response structure, ensuring that its resonant frequency covers the various dynamic motion frequencies that the human body may experience during normal cycling, for example, forming a dense frequency response band within the typical range of 0.5 to 10 Hz. This setting ensures that when the raw sensory signal contains a component close to a specific resonant frequency, the corresponding resonant branch will resonate and significantly amplify that component, thereby enhancing the frequency information related to human motion behavior in the signal while effectively suppressing interference from background noise and other irrelevant frequency bands, resulting in a clearer output signal in terms of amplitude and more selective spectrum. The resonantly amplified sensory signal is then introduced into a high-fidelity multimodal physical mixer circuit for joint feature extraction. This mixer integrates multiple analog adder and multiplier components, capable of cross-calculating, nonlinearly combining, and reconstructing the signals from different sensor sources in the time domain. This hybrid transformation allows the original local signal to express the overall spatial dynamic distribution. During this process, the mixer outputs a set of integrated energy distribution signal data based on the mutual coupling between multiple sensing channels, forming a comprehensive characteristic signal that describes the spatial transfer trend of weight on the pedals. This signal combines multiple sources, has a clear direction, and can be analyzed for its changing trend, effectively reflecting the rider's intended movements on the scooter. The comprehensive characteristic signal is input into a phase-locked amplifier circuit module, where the phase-locked mechanism extracts and enhances the effective components of the signal. A reference signal source, provided by a human motion model or the system's inertial measurement unit, is introduced into the circuit. This signal is compared with the input somatosensory characteristic signal, and the degree of phase synchronization between the two is determined by calculating the phase ratio. When the phase of the signal is consistent with the reference source, it indicates that the signal is behaviorally relevant, and the system amplifies it with a higher gain. Random perturbations with inconsistent phases or large frequency drift are automatically attenuated due to the selectivity of the phase-locked mechanism, resulting in a somatosensory signal with a high signal-to-noise ratio, strong dynamic correlation, and stable phase. Based on the enhanced signal with a high signal-to-noise ratio, a spatiotemporal gradient analysis is performed. By analyzing the direction of change and instantaneous increase and decrease trends on the sensor array, the main direction of the cyclist's weight distribution and the acceleration characteristics during the movement force are extracted. This information together forms a somatosensory characteristic signal with clear physical meaning, strong timeliness, and can reflect the actual behavioral intention.

[0020] S3, based on the somatosensory characteristic signals, a dynamic model of the scooter-rider coupling system is established, and real-time state calculation is performed to obtain the current state parameters and state deviation values; It should be noted that the high signal-to-noise ratio spatial distribution data contained in the somatosensory characteristic signals is used to extract the coordinates of the rider's actual point of action on the pedals, the center of gravity offset represented by the pressure distribution characteristics, and the dynamic body posture parameters inferred from the micro-acceleration response curve and high-frequency perturbation trends. Through the joint processing of this information, a state observation dataset containing multidimensional spatiotemporal variables is formed. This dataset reflects the interactive behavior characteristics of the human body on the scooter. Combined with the scooter's structural parameters such as wheelbase, pedal geometry, vehicle mass, and center of mass position, a preliminary correlation mapping between the vehicle and the human body is completed, establishing the state input foundation for system-level modeling. On this basis, in order to accurately express the dynamic coupling behavior between the scooter and the rider, a system configuration description matrix was constructed. This matrix represents the spatial state of the entire system with multiple generalized variables describing the position, orientation, and joint state. Its motion law is derived through the Lagrangian dynamics method. In this law, the inertial characteristics of the system are reflected by the mass matrix, the centrifugal and interaction caused by speed changes are characterized by the Coriolis force term, and the gravitational influence on the system in different postures is expressed by the gravity term. The entire dynamic model integrates the flexible structure of the human body, the rigid structure of the scooter, and the physical coupling between them generated by the contact interface, forming a continuous, controllable dynamic response structure suitable for feedback loops. A mirror state estimator is constructed based on the dynamic model of the scooter-rider coupled system. This estimator uses the extended Kalman filter algorithm as its core calculation mechanism. During the state prediction phase, the constructed dynamic model is used to extrapolate the current system state over time, and the uncertainty range of the system's future state is calculated based on the model structure. During the state update phase, the observation vector is constructed using triaxial acceleration and angular velocity data from the inertial measurement unit (IMU) and key parameters extracted from the somatosensory characteristic signals. The predicted output is compared with the actual observations using a measurement function to obtain the state error. This error is then used to iteratively correct the system state and error covariance to obtain the current state parameters, which include key information such as position, attitude, velocity, acceleration, and contact force feedback. By comparing the current estimated state with a preset ideal riding model, a reference state variable representing the ideal control behavior is constructed. The difference between the two is then calculated to obtain the state deviation. By weighting each variable according to its importance within the control objective, an overall deviation value is calculated to represent the degree of error between the scooter's actual response and the target behavior.

[0021] Based on somatosensory characteristic signals, a state observation dataset reflecting the scooter-human coupling behavior was constructed. This dataset was used to extract motion configuration information, including the position parameters of the scooter's center of mass in the ground coordinate system, the vehicle's body orientation angle, and the rider's upper and lower limb joint postures and relative limb angles. By uniformly encoding and normalizing these multi-source variables, a system configuration description matrix with clear physical correspondences was formed. This matrix, expanded by generalized state variables, can characterize the absolute motion state of the scooter body within a plane and capture the relative changes in human motion relative to the scooter. Based on the configuration description matrix, mass attribute modeling was performed for each system component, specifically the rigid components of the scooter body (such as the pedals, frame, and wheels) and the mechanical equivalent modeling of the rider's body on the pedals. The analysis considers the inertial variations caused by the uneven distribution of human mass between the feet. A local mass model is constructed based on the known structural parameters of the scooter. These models integrate the positions and masses of each mass unit, as well as their contributions to the overall system's inertial response, to form the system's mass matrix. This matrix describes the system's inertial propagation structure and response amplitude to external forces in any configuration. Based on this mass matrix and configuration description matrix, the interactions between velocity variables in the system are analyzed. Specifically, when the scooter experiences dynamic behaviors such as turning, acceleration, and forward and backward swaying, coupling occurs between velocities and angular velocities in different directions. These coupling effects manifest as Coriolis inertial forces in the dynamic response. By applying partial derivative approximations and posture rotation operations to the time-varying characteristics of the generalized velocity variables, the interactions between the velocity components are extracted, and a Coriolis force matrix is constructed. This matrix reflects the inherent dynamic interference characteristics of the system caused by the asynchrony of velocity propagation paths during non-uniform motion. Based on the spatial structure of the configuration description matrix, a gravitational potential energy analysis is performed. By integrating the vertical projection heights of the scooter and human mass units, the overall gravitational potential energy function of the system is derived. By solving the rate of change of this function in state space relative to the various configuration variables, a set of gravity term vectors is generated that represents the tendency of different components to be affected by gravity. By combining and integrating these three substructures—the system mass matrix, the Coriolis force matrix, and the gravity term—a complete coupled dynamic model of the scooter and rider is constructed.

[0022] Based on the dynamics model of the scooter-rider coupled system, a state-space representation that conforms to the model's physical constraints is constructed. The system state variables are organized into a complete state vector. This state vector contains the position coordinates of the scooter's center of mass, its direction of motion, its body attitude angles, the attitude angles of key body nodes of the rider, and their first-order derivatives with respect to time, namely, linear velocity, angular velocity, and attitude rate of change. Furthermore, a state transition function is established to describe the dynamic evolution of the system state over time. A measurement function is constructed, combining sensor observables, to map the state variables to observable outputs measured by the body sensors and IMU, thus forming the complete structure of the state-space model. Within the state estimator, a temporal update process is performed on the state vector based on this state-space description. Using the system's dynamics model, the state is advanced at the current known state point to obtain the predicted state data for the next time step. Furthermore, based on the model's internal uncertainty and the cumulative error trend of state propagation, a prediction covariance matrix is calculated to quantify the uncertainty range of the predicted state, thereby forming a preliminary estimate of the future state. High-precision somatosensory characteristic signals and measured data from an inertial measurement unit are fed into the estimator as external observation inputs. These are then compared with the predicted state data using a measurement function to calculate predicted observations. These are then subtracted from the actual observations to form a set of observation residuals, representing the deviation between the current model prediction and the sensor data. To perform weighted correction for these deviations, the measurement noise matrix and the prediction covariance matrix are combined, and a set of Kalman gain coefficients is calculated using a derivation mechanism based on Kalman filtering theory. These gain coefficients reflect the relative weight between the model prediction reliability and the measurement data reliability at the current moment. These gain coefficients are then applied to the predicted state data as state correction factors, and a state update operation is performed to obtain the correction for the current state. Based on this, the system state covariance matrix is updated to ensure that the corrected uncertainty diffusion trend is accurately reflected in the next prediction phase. Ultimately, through an iterative cycle of dynamic prediction and measurement updates, the mirror state estimator achieves real-time state tracking and estimation output for the scooter and rider during complex motion, ultimately obtaining the current state parameters.

[0023] S4, performing a somatosensory weight distribution change mapping conversion on the current state parameter and the state deviation value to obtain a riding control demand parameter; Specifically, the system analyzes the changing trends of the somatosensory inputs implied by the current state parameters to identify the rider's weight distribution pattern on the pedal surface. By performing a time-series analysis of the pressure response at each location in the sensor array, the rider's dynamic center of gravity trajectory is restored. Combined with the system's output of velocity, acceleration, and posture information, key behavioral features such as forward lean, backward lean, and left-right deviation are identified. The current weight distribution components in the fore-aft and left-right directions are then extracted and numerically differentiated as a rate of change. This results in a weight distribution change vector containing both directional and dynamic amplitude information, reflecting the direction and intensity of the rider's active control intent. The weight distribution change vector is then processed jointly with the system state deviation value. A set of error function models are introduced to perform differential analysis, establishing a mapping residual field. This residual field can be mathematically converted into a control deviation matrix, which represents the degree of inconsistency between the current somatosensory input and the ideal riding target. This control deviation matrix includes directional deviation errors and reflects issues such as dynamic response delay and amplitude response mismatch. The control deviation matrix is deconstructed and calculated using the control law. Combining the system's control constraints with the actual actuator's physical capabilities, the matrix elements are parsed into mechanical command parameters that can actually act on the scooter. The longitudinal component is converted into a linear driving force, while the lateral offset is mapped into a lateral steering force. This results in a set of basic drive demand data, representing the physical action requirements for the scooter's current response and forming the first instruction layer for system output regulation. This basic drive demand data is then combined with the scooter's current operating state for calculation. The Jacobian matrix is introduced as a state mapping operator in this process. By modeling the differential relationship between kinematic variables, an instantaneous linear approximation model from control input to system response is established. This Jacobian matrix is then integrated with a proportional-differential controller. The proportional control term adjusts the current output response amplitude, while the differential term suppresses high-frequency fluctuations during the change process, thereby refining the control command and ultimately generating the riding control demand parameters.

[0024] S5, calculating a real-time drive control signal sequence according to the riding control requirement parameters, and dynamically allocating the front and rear wheel motor torques to the real-time drive control signal sequence to obtain a target motor drive execution instruction set.

[0025] The riding control demand parameters are input into a fractional-order elastic controller. This controller, based on fractional-order calculus theory, focuses on the introduction of non-integer-order differential and integral operators, providing a flexible control mechanism for the scooter that lies between traditional PID control and model predictive control. By setting the order parameters of the fractional-order differential and integral operators, the system's response sensitivity to high-frequency disturbances or low-frequency trend changes is adjusted based on the dynamic input behavior of the rider. Simultaneously, by adjusting the proportional, integral, and differential gain coefficients, the controller achieves an adjustable balance between amplitude, rate, and hysteresis control. The controller outputs a set of fractional-order control signals that combine immediate state response with historical error memory. These signals exhibit scalable memory characteristics and nonlinear dynamic adjustment capabilities. Based on the fractional-order control signals, an elastic response control command is established that lies between speed control and position control, balancing the stability of position tracking with the real-time performance of speed adjustment. Dynamic interpolation is used to express the compliance relationship with changes in rider input. To ensure the stability and safety of the control process within the energy range and dynamic delay, a Lyapunov energy function is introduced to perform constraint analysis on the control signal. This involves calculating the total kinetic and potential energy generated by the system's current motion state in real time and setting an upper energy threshold. When a control command could cause the system's energy to exceed the set limit, a gradient projection method is used to constrain the amplitude and direction of the control signal, ensuring that the motor drive does not overload or deflect abnormally. Furthermore, a time-delayed feedforward prediction model, taking into account the inherent delays in the system's sensor and motor responses, is used to compensate for the impending control state. This ensures that the generated real-time drive control signal sequence is consistent with the actual response of the physical system in terms of timing, avoiding unstable behavior caused by system control advances or lags. Dynamic torque distribution between the front and rear wheel motors is then applied to this drive signal sequence based on the scooter's current operating state and external road adhesion conditions. The torque distribution strategy considers factors such as changes in the ground friction coefficient, tire load distribution, and the vehicle's turning radius, assesses the difference in traction between the front and rear wheels in real time, and adjusts the torque distribution based on the decomposition of the longitudinal and lateral control variables. On high-adhesion roads or when running in a straight line at high speed, the front-wheel drive component is prioritized to enhance traction; in scenarios with low adhesion or frequent steering movements, more torque is appropriately allocated to the rear wheels to enhance lateral stability, ultimately forming a set of target motor drive execution instructions with strong targeting and fast dynamic response.

[0026] The system analyzes the adhesion characteristics of the scooter's current road surface in real time. By integrating longitudinal acceleration, wheel speed curves, wheel slip levels, and subtle side slip signals from the sensor, it extracts an adhesion index reflecting the friction between the wheels and the road surface. This adhesion index dynamically quantifies the impact of different road conditions (such as dry asphalt, wet concrete, gravel, or sand) on the efficiency of driving force transmission. It also combines wheel speed and acceleration data to estimate the actual adhesion limit of the wheels. This allows it to establish a distribution model for the adhesion of the front and rear wheels under the current road conditions. Based on this model, it calculates the front and rear wheel torque distribution coefficients, enabling the drive system to prioritize torque output to the wheels with greater adhesion, avoiding slippage and energy waste in low-friction environments. After obtaining the front and rear wheel torque distribution coefficients, the real-time drive control signal sequence is vector-decomposed, decomposing the vehicle's control commands into longitudinal traction control components and lateral steering force control components. A steering coefficient is introduced based on the system's preset steering structure parameters to describe the actual sharing of the front and rear wheels during lateral response. A torque combination model is then constructed based on this steering coefficient and the front and rear wheel distribution coefficients to calculate the torque commands for the front and rear wheels, respectively. This calculation process ensures that the sum of the dynamic components in each direction in the torque space satisfies the vehicle's dynamic balance principles while also providing pre-adjustment and buffering for sudden maneuvers, enhancing the control system's dynamic flexibility. To prevent structural oscillations, energy peaks, or riding discomfort caused by sudden torque changes, a torque rate-of-change limiting mechanism is applied to each of the calculated front and rear wheel torque commands. By comparing the difference between the current torque command and the output value of the previous control cycle, the torque command is clipped within a set maximum rate-of-change range to achieve a smooth transition torque output value. The smoothly transitioned torque output value is compared and analyzed with the real-time safety status model of the scooter. The model is constructed based on parameters such as speed, posture stability, vehicle body tilt angle, gyro rate change, and historical fault records. When the system detects that any key safety indicator is approaching the risk boundary, the control mode switching mechanism is triggered. By adopting a continuous control function to perform weighted transition processing on the normal control output and the safety control output, the system control logic smoothly transitions to a conservative control mode under restricted conditions, thereby ensuring that the controllability and basic stability of the vehicle motion can still be maintained under extreme conditions, and generating a target motor drive execution instruction set that conforms to real-time road conditions, adapts to the current state, and takes into account safety redundancy.

[0027] In one example, collecting raw somatosensory signals from a scooter includes: A micro-accelerometer array is installed on the pedal surface of the scooter, pressure sensors are set on both sides of the pedal, and a six-axis inertial measurement unit is assembled at the central control unit to obtain a somatosensory data acquisition network; The rider's weight distribution change data and scooter posture data are acquired in real time through the body sensing data acquisition network to obtain the original sensor signal.

[0028] In this example, a micro-accelerometer array is installed on the scooter's pedal surface. This array adopts a triangular lattice structure to ensure uniform coverage and orientation independence in a two-dimensional plane. In the actual deployment, 12 micro-accelerometers are selected as the basic unit, with a spacing of 5 cm between sensors. These accelerometers have extremely high sensitivity, with a minimum measurement resolution of 0.001g. They feature a sampling rate of up to 1000Hz, ensuring stable data output even in the presence of rapid changes in physical sensation or vibration interference. They also have a dynamic measurement range of ±2g to accommodate the acceleration characteristics of various riding movements. Pressure sensors are also deployed on both sides of the pedal to enhance the system's ability to detect pressure differences between the rider's left and right feet. These pressure sensors feature a high sensitivity of 0.05N and a wide range of measurement, covering 0 to 1000N from static standing to rapid compression, enabling them to capture subtle changes in foot load under different postures and maneuvers. By comparing the output signals of the two lateral pressure sensors, the balance of the rider's foot load center can be determined and used to infer the degree of body roll during steering maneuvers, providing important signal support for lateral posture adjustment. The central control unit integrates a high-performance six-axis inertial measurement unit (IMU), consisting of a three-axis accelerometer and a three-axis gyroscope. This unit monitors the scooter's overall spatial motion, including key dynamic characteristics such as translational acceleration and rotational angular velocity. The three-axis accelerometer senses the scooter's linear acceleration components in real time in the x, y, and z directions, while the gyroscope acquires rotational angular velocity information about the three axes, providing vehicle posture recognition capabilities. The IMU's signal output frequency is also required to be synchronized with the micro-acceleration array. A central synchronous clock management module ensures the time consistency of all sensor data for subsequent spatial posture fusion and body behavior decoding. All these sensor units are connected to the central signal conditioning circuit via differential signal cables and are electromagnetically shielded to resist external electromagnetic interference. The signal conditioning circuit provides both primary amplification and bandpass filtering. It pre-amplifies the sensor's weak electrical signals to a standard level range for easy analog-to-digital conversion. It also eliminates high-frequency jitter and low-frequency drift through filtering, significantly improving signal validity. The conditioned signals are transmitted uniformly to the central processing unit, which performs time alignment, format standardization, and batch integration of data from different sensor channels to form a standardized raw somatosensory data stream. As the data stream is aggregated in the central processing unit, the system captures real-time data on changes in the rider's weight distribution on the pedals and the scooter's own posture status.Weight distribution change data is obtained by reversely analyzing the pressure-induced microvibration response of the microaccelerometer array, extracting the direction and magnitude of acceleration change per unit time at each point. The spatial relationship between the sensing points is then used to calculate the trajectory of the rider's plantar pressure center within the pedal plane. Continuity analysis and rate-of-change assessment of this trajectory determine whether the rider is in a stable standing position, gradually shifting their center of gravity in preparation for acceleration, or experiencing a sudden shift and impending a turn. The scooter's attitude data is provided by the IMU. Its acceleration and gyroscope signals are filtered and fused using an attitude calculation algorithm to obtain absolute attitude parameters such as the scooter's roll, pitch, and yaw angles. These parameters are then paired with the somatosensory data to simultaneously model the scooter's dynamic behavior and the rider's control actions. This multi-channel, multi-modal fusion of raw signals forms the input layer of the scooter's somatosensory control system, ultimately resulting in the raw sensor signals.

[0029] In one example, the original body sensory signal is subjected to resonant amplification and multimodal mixing to obtain a body sensory characteristic signal, including: The original body sensory signal is input into the LC resonant circuit array. By setting the inductance and capacitance parameters of the LC resonant circuit array, the resonant frequency is matched with the characteristic frequency range of human motion to obtain a resonantly amplified body sensory signal. The multimodal physical mixing circuit combines the resonantly amplified body sensory signals to obtain a comprehensive characteristic signal of weight distribution changes; The comprehensive characteristic signal of the weight distribution change is input into the phase-locked amplifier circuit, the phase ratio between the reference signal in the phase-locked amplifier circuit and the comprehensive characteristic signal of the weight distribution change is calculated, and the effective signal in the comprehensive characteristic signal of the weight distribution change is amplified according to the phase ratio to obtain a body sensing signal with a high signal-to-noise ratio; The direction and acceleration information of weight distribution changes are extracted based on the high signal-to-noise ratio somatosensory signal to obtain the somatosensory feature signal.

[0030] In this example, the raw sensory signal is fed into an LC resonant circuit array. Each pair of inductors and capacitors forms a resonant unit at a specific frequency. These units are arranged in an array to form a filter-amplifier structure with a frequency distribution covering 0.5 to 10 Hz, thereby enhancing the physical characteristics of various dynamic sensory behaviors experienced by humans on a scooter. By adjusting the inductor and capacitor values in the LC circuit, the resonant frequencies of each resonant branch correspond to the frequency characteristics of different human behaviors, such as posture changes, weight shifts, turning preparations, and pedaling frequency. When the raw signal contains a frequency component close to the frequency of a resonant branch, a local resonance reaction occurs in that branch, significantly amplifying the amplitude of that frequency component at the output, effectively enhancing the target signal. Background noise outside the resonant frequency range is effectively attenuated due to impedance mismatch. This signal enhancement mechanism, achieved through "frequency locking," exhibits selective amplification, completing a pre-filtering and feature enhancement operation based on the spectral distribution of human motion at the circuit level. The output signal constitutes a resonantly amplified sensory response with enhanced directionality and energy characteristics. The resonantly amplified signal is introduced into a multimodal physical mixing circuit for advanced cross-fusion and feature construction. This mixing circuit is equipped with multiple high-precision analog adders and multipliers. Its operational structure allows for the combination of enhanced signals from different frequencies, spatial locations, and sources at the circuit level. The mixer adds and multiplies the multiple resonantly amplified input signals, generating high-order coupled signals with spatially differentiated distributions and temporally synchronized variations. This method preserves the energy spectrum characteristics of the original signal and constructs a composite signal that reflects the overall weight transfer trend and the synchronized behavior of both legs. This composite characteristic signal of weight distribution changes is then input into a phase-locked amplifier circuit. The basic principle of a phase-locked amplifier is to phase-match the input signal with a reference signal of the same frequency. The phase difference between the two signals is continuously monitored to determine whether the input signal is the expected valid component of the system. The reference signal is set internally by the system or derived from the dynamic behavior frequency provided by the IMU. When the input signal matches the reference signal in phase, the phase-locked amplifier circuit outputs a synchronized signal with significantly enhanced amplitude and suppresses phase-inconsistent high-frequency jitter or low-frequency interference, resulting in a high signal-to-noise ratio (SNR) somatosensory signal. Directional and acceleration characteristics are extracted from this high-SNR signal, relying primarily on the spatial topology of the sensor array. By comparing the response time differences, energy amplitude differences, and rate of change differences between different sensor points, and using a discrete gradient operator to perform differential operations on the array signals, the real-time trajectory of the rider's foot pressure center of gravity on the pedal surface is restored. The direction, velocity, and acceleration of the center of gravity change are then determined by differentiating the trajectory.If the trajectory shows a stable, unidirectional increase, it indicates that the rider is accelerating and leaning forward; if the trajectory swings rapidly from side to side, it indicates turning preparation; if the trajectory is concentrated at the rear and the amplitude decreases, it indicates deceleration or braking. These judgments are extracted by extracting the dynamic characteristics of the signal in time and space, forming a set of somatosensory characteristic signals that are used to drive the control logic.

[0031] In one example, a scooter-rider coupled system dynamics model is established based on somatosensory characteristic signals, and real-time state calculation is performed to obtain current state parameters and state deviation values, including: The rider's position coordinates, center of gravity distribution, and posture data on the pedals are extracted based on the somatosensory feature signals, and combined with the scooter's body parameters to obtain a state observation dataset. Based on the state observation dataset, the system configuration description matrix of the scooter and rider is constructed and the scooter-rider coupled system dynamics model including the mass matrix, Coriolis force matrix and gravity term is derived. A mirror state estimator is constructed based on the dynamic model of the scooter-rider coupling system. The somatosensory characteristic signals and inertial measurement unit data are input into the mirror state estimator. The state is predicted and updated through the extended Kalman filter algorithm to obtain the current state parameters. The state deviation value between the ideal state and the current state is calculated based on the current state parameters and the preset ideal riding model.

[0032] In this example, high-signal-to-noise ratio somatosensory characteristic signals are used as the analysis basis. The rider's position on the pedal surface is recovered from the pressure response pattern of a multi-point sensor array. The center of gravity position coordinates are constructed using the instantaneous response amplitude and distribution gradient of each sensor unit. The center of gravity motion trajectory is then inferred based on the time series trend. The rider's posture change trend is then estimated based on the speed changes, amplitude mutations, and left-right deviation patterns in these trajectories. To ensure the physical interpretability of the extracted results, scooter parameters are introduced, including the pedal geometry, overall mass distribution, front and rear wheelbase, body rigidity and inertial configuration, etc. The human-machine interface is used as a coordinate mapping basis. This allows the extraction results to go beyond the two-dimensional surface perception layer and be projected into the system coordinate space in conjunction with the scooter's physical geometry. This results in a state observation dataset that simultaneously expresses the rider's behavioral characteristics and the geometric semantics of the entire vehicle. Based on this state observation dataset, a system configuration description matrix is constructed to characterize the overall motion behavior of the scooter-rider combination. This matrix, organized around generalized coordinates, uniformly encodes parameters such as the scooter's center of mass, body attitude angles, tire states, and key rider posture variables (such as foot plantar angles, lower limb joint inclinations, and trunk center of gravity height) into a configuration space. This allows for a unified representation of the complex human-machine system's multi-degree-of-freedom, multi-level, and multi-source states. Furthermore, Lagrangian dynamics theory is employed to model the system's inertial properties, energy transfer paths, and control influences during arbitrary state changes based on the principle of least action. Structural parameters describing the system's dynamic response are derived, including a mass matrix representing the overall mass distribution, a Coriolis force matrix characterizing the system's velocity coupling, and a gravity term describing the distribution of the system's gravitational potential energy. These structural mechanical terms are determined based on the distribution of configuration variables in state space and exhibit significant geometric and state dependence. They can adaptively represent the complex dynamic coupling between rider and scooter under varying riding behaviors. By integrating these mechanical parameters, a scooter-rider coupled dynamics model is constructed, providing a continuous, differentiable, and stable response description function for practical control. To connect the theoretical modeling described above with real-time control requirements, a mirror state estimator was constructed based on this dynamic model. This estimator uses system dynamics as its prediction core and sensor observations as its correction basis. It uses an extended Kalman filter algorithm to iteratively integrate prediction and update. In actual operation, the system's current state vector and control input are fed into the dynamic prediction function to obtain the state estimate and error covariance for the next moment. Then, somatosensory characteristic signals and observations such as linear acceleration and angular velocity provided by the inertial measurement unit are introduced to construct the difference residual between the actual observation and the predicted value. A set of dynamically adjusted gain coefficients are calculated based on the credibility of the observation signals and the system uncertainty to correct the predicted state.Because scooter motion is accompanied by highly nonlinear perturbations such as sudden acceleration changes, rider posture deviations, and lateral inertial impacts, the extended Kalman filter's nonlinear processing capabilities ensure that state estimation convergence and response accuracy are maintained even in highly dynamic environments. The current state parameters obtained after filtering include the scooter's precise posture position, velocity state, and acceleration trend, as well as the rider's center of gravity dynamics and motion deviation amplitude. These current state parameters are then compared and analyzed with a set of preset ideal riding models. These ideal models, constructed from typical trajectory curves, stable posture structures, or empirical control strategies, represent the state behavior that the system should exhibit under ideal conditions. The state deviation value is calculated by comparing the current state vector with the ideal state vector and performing a weighted combination of the difference between the two.

[0033] In one example, based on a state observation dataset, a system configuration description matrix of the scooter and rider is constructed, and a scooter-rider coupled system dynamics model containing a mass matrix, a Coriolis force matrix, and a gravity term is derived, including: Based on the state observation dataset, the scooter's center of mass position, orientation angle, and rider's posture angle are combined into a system configuration description matrix; Calculate the mass attributes of the system configuration description matrix, analyze the mass distribution and inertia parameters of each part of the scooter and rider, and obtain the system mass matrix; The velocity coupling characteristics are calculated based on the system configuration description matrix and the system mass matrix, and the Coriolis force matrix is constructed by calculating the velocity derivative and angular velocity. The gravitational potential energy of the system configuration description matrix is analyzed, the gravity term is calculated, and the system mass matrix, Coriolis force matrix and gravity term are combined to construct a scooter-rider coupled system dynamics model.

[0034] In this example, the scooter's center of mass position in the global reference frame is extracted from the state observation dataset. This includes horizontal coordinate components and vertical center of gravity height variations, as well as its orientation angle information, which is used to characterize the vehicle's rotational posture in the three degrees of freedom (Pitch, Roll, and Yaw). Furthermore, the rider's pedaling motion is calculated, specifically the stance deviation caused by plantar pressure distribution, pelvic posture changes, and center of gravity tilt caused by upper limb coordination. These are encoded as attitude angle information. All these variables are then reorganized into a temporally and structurally consistent configuration variable set, forming a configuration description matrix for the scooter-rider system at the current moment. This matrix, arranged in generalized coordinates, includes both displacement and attitude variables, as well as derivable angular and linear velocity variables. Based on this configuration description matrix, mass attribute mapping models are established for the scooter and individual riders, respectively, to form a mass matrix that reflects the system's inertial response capability. For the scooter, the mass distribution characteristics of components such as the frame, pedals, battery, motor, and front and rear wheels are considered. The mass, center of mass position, and moment of inertia of each component are modeled based on their position, shape, and connection method in the vehicle coordinate system. For the rider, the rider is approximated as a rigid body structure with multiple flexible joints. Standard biomechanical parameters are used to estimate the mass proportions and corresponding inertia distribution characteristics of each body part. The feet, legs, torso, and head are mapped to the vehicle coordinate system based on the posture angles reflected by sensor data, thus achieving a unified representation of the mass distribution of the entire human-vehicle coupled system. After completing the above mass data projection, the mass contributions of each scooter component and each part of the rider are integrated into the total system mass matrix through inertia integration. This matrix provides a fundamental representation of the dynamic response of the entire system under arbitrary acceleration input and directly influences the adjustment speed and execution inertia of subsequent control commands to change motion. To describe the dynamic effects of the scooter's speed coupling in different configurations, the relationship between velocity and angular velocity variables is analyzed when obtaining the mass matrix. Based on the joint calculation of the system's velocity derivatives, angular velocity, and attitude transformation matrices, the inertial interference term, known as the Coriolis effect, caused by the nonlinear superposition of velocity propagation paths between different degrees of freedom is extracted. In constructing the Coriolis force matrix, the partial derivative relationship between generalized coordinates is considered to analyze whether the scooter's roll affects the rider's longitudinal stability or whether it induces involuntary reverse posture adjustments in the rider's upper limbs during acceleration. This nonlinear velocity interference will manifest in the Coriolis term as a coupled velocity term, whose structural characteristics continuously adjust as the system state changes. Therefore, it is expressed as a configuration-dependent unsteady matrix to ensure that the system's true inertial coupling strength is reflected at different state space locations, thereby ensuring higher dynamic self-consistency and physical accuracy of the model.To construct the system's true energy structure under the influence of a gravitational field, a gravitational potential energy mapping analysis is performed on the configuration description matrix. This involves establishing a gravitational potential energy expression relationship between the height position of each mass unit and its orientation to the ground. The gravitational forces acting on all mass units in their current posture are then summed to construct a gravity term vector that reflects the system's vertical posture stability and the influence of gravity on its response trend. This term, along with the mass matrix and the Coriolis matrix, forms the scooter-rider coupled dynamic equation. Combining these three types of mechanical structural terms creates a nonlinear dynamic model that describes the system's ability to respond to input forces or torques in any configuration. This model boasts adjustable structure, coherent variables, and well-defined parameter coupling. It can not only describe system stability under static equilibrium conditions, but also characterize inertial effects, coupled interference, and stability-destroying trends in dynamic scenarios such as high-speed driving, sharp turns, or sudden load changes, providing a physically consistent foundation for state estimation, path prediction, and control strategies.

[0035] In one example, a mirror state estimator is constructed based on the dynamic model of the scooter-rider coupling system. The somatosensory characteristic signals and inertial measurement unit data are input into the mirror state estimator. The state prediction and measurement update are performed using the extended Kalman filter algorithm to obtain the current state parameters, including: Based on the dynamics model of the scooter-rider coupled system, a mirror state estimator is constructed. The system state variables are organized into a state vector containing position, angle, and their derivatives. The state transfer function and measurement function are set to obtain a state space description. Perform time update processing on the state vector in the state space description, and calculate the state prediction value and prediction covariance matrix according to the system dynamics equation to obtain preliminary predicted state data; Taking the somatosensory characteristic signals and inertial measurement unit data as observation inputs, the deviation between the preliminary predicted state data and the actual measured value is calculated, and the Kalman gain coefficient is calculated based on the measurement noise matrix and the predicted covariance matrix to obtain the state correction factor; The state correction factor is applied to correct the preliminary predicted state data, and the system state covariance matrix is updated to obtain the current state parameters.

[0036] In this example, based on the dynamics model of the scooter-rider coupled system, the generalized coordinate system and kinematic variable structure established during the modeling process are converted into the state-space representation required for state estimation. During this process, the composition of the system state variables is clarified, including the two-dimensional or three-dimensional spatial position of the scooter's center of mass, the heading angle, roll angle, and pitch angle of the scooter, as well as the time derivatives of these positions and angles, namely, linear velocity, angular velocity, and acceleration. Furthermore, the derivatives of the rider's center of gravity, the position of the plantar pressure center, and the angle of posture change are incorporated into the state variable set, forming a high-dimensional state vector that comprehensively reflects the current physical state of the human-machine coupled system. After constructing the state vector, the state transition function is derived based on the dynamics model. This function describes the physical evolution of the system from the current state to the next state under given input conditions. This function is derived based on the mass distribution, velocity coupling characteristics, and gravity response behavior of the coupled system, while preserving the time correlation of nonlinear terms and the cross-correlations between variables. A measurement function is also defined to project the system's internal state variables into observable space. This mapping logic from state variables to sensor readings is constructed using the dimensions captured by the somatosensory signature and IMU data as a reference. This measurement function includes an estimation method for the scooter's pitch angle and the rider's center of gravity offset, effectively linking highly abstract state information with real-world physical measurements. A time-based update process is performed on the state vector in the state space description. At each cycle, the state variables for the next moment are predicted based on the current state estimate and the transfer function. The state covariance prediction is calculated based on the relationship between velocity derivatives and angular momentum in the dynamics model, forming a preliminary predicted state. To correct this predicted state, the somatosensory signature and IMU inertial measurement data are used as actual observation inputs during each update cycle. The predicted observation value is calculated using the measurement function and compared with the actual measurement value to obtain the observation residual. This residual represents the error between the prediction model and the actual observation and serves as an important basis for dynamically adjusting the state estimate. The Kalman gain coefficient is calculated by combining the current predicted state covariance matrix with the sensor measurement noise matrix. This gain value reflects the weighted relationship between the uncertainty of the model prediction and the reliability of the measurement results. When the measurement error is small and the prediction error is large, the system tends to rely on the measurement results for correction; when the measurement data has large jitter, it tends to maintain the model prediction. The obtained Kalman gain is multiplied by the observation residual and applied to the predicted state vector, completing the state correction process. This correction operation can appropriately adjust the direction and magnitude of the prediction offset, making the final state estimate closer to the actual system behavior. At the same time, based on the current gain adjustment result, the state covariance matrix is updated to reflect the corrected uncertainty propagation trend, thereby providing a more realistic covariance initial value for the next cycle's state prediction.Through the continuous iterative operation of this mirror state estimator, the scooter control system can continuously and stably generate real-time estimation results of the current state parameters in the face of nonlinear interference, complex motion input and multi-source sensor fusion.

[0037] In one example, a somatosensory weight distribution change mapping conversion is performed on the current state parameters and the state deviation value to obtain the riding control demand parameters, including: Analyzing the change pattern of the rider's weight distribution based on the current state parameters, determining the center of gravity offset and change rate in the front, back, left, and right directions based on the change pattern, and constructing a weight distribution change vector based on the center of gravity offset and change rate; The error function is calculated for the weight distribution change vector and the state deviation value to obtain the control deviation matrix; Deconstruct the control law calculation of the control deviation matrix and convert the control deviation matrix into the required driving force and steering force to obtain the basic driving demand data; According to the basic driving demand data and the current operating status of the scooter, the Jacobian matrix is calculated and combined with the proportional-derivative controller to generate the riding control demand parameters.

[0038] In this example, the dynamic coupling between the rider and the scooter is extracted based on the system's current state parameters, and the changing weight distribution pattern on the pedal surface is analyzed. This analysis, based on contact pressure and acceleration data provided by the sensor array, reconstructs the trajectory of the rider's foot's center of force within the pedal plane by differentiating and fitting the response amplitudes of each sensor point over a time series. This trajectory is then mapped into fore-aft and lateral displacement paths. Combined with the scooter's posture data and motion angles provided by the inertial measurement unit, the planar coordinates of the foot's center of force are projected into the scooter's coordinate system. This eliminates artifacts caused by vehicle motion and ensures that the center of gravity offset reflects only the rider's true control behavior. Once the center of gravity offset trajectory forms a continuous curve in time, its first-order derivative is processed to obtain the rate of change. This is then combined with the acceleration difference between the current and previous frames to construct the center of gravity change trend, thereby simultaneously obtaining the rider's instantaneous displacement and rate of change in the fore-aft, lateral, and left-right directions. On this basis, these components are combined in spatial order to form a four-dimensional weight distribution change vector. This vector provides directional characteristics of the rider's movements and reflects the continuity and abruptness of their control actions. This weight distribution change vector is then combined with the system state deviation value obtained by the state estimator. The difference between the two is quantified by designing an error function. This error function should comprehensively consider three factors: movement direction consistency, response amplitude proportionality, and change rate synchronization. This establishes a deviation expression structure with directional coupling characteristics. Mathematically, this structure is based on vector differences. The dynamic weights of the error terms are used to construct a control deviation matrix. This matrix contains multidimensional mismatch information between the current somatosensory input and the system response. The control deviation matrix is deconstructed and transformed. The error information of different dimensions in the matrix is mapped into physically meaningful output targets, namely longitudinal driving force and lateral steering force, through a specific control law. During this process, the control law is formulated based on the scooter's physical structure and dynamic model. This control law identifies the dominant component in the deviation matrix and assigns it a higher control weight, while also damping and limiting small error terms in non-dominant directions. This ensures convergence, stability, and enforceability of the output control variable. After this processing, the basic drive demand data is output—a set of quantized instructions describing the linear thrust and angular deflection force required by the scooter at the current moment. To improve the physical adaptability and dynamic response stability of the control instructions during execution, a Jacobian matrix is established based on the scooter's current operating state. This matrix describes the differential mapping between the control input and the vehicle's actual response. By introducing the Jacobian matrix, the basic drive demand data is projected into a state change rate under the current system state, ensuring that the control instructions remain valid and physically consistent even in the presence of structural nonlinearities or inertial changes.This mapping result is then combined with a proportional-derivative controller to enhance the system's feedback regulation capabilities. The proportional component corrects the current error amplitude, while the differential component predicts future system trends and proactively suppresses sudden changes. The combination of these two creates a control output that is both responsive and stable. After these calculations and compensations, the desired riding control parameters are output. This set of parameters reflects the scooter's current core control objectives, including direction of motion, speed adjustment, and posture balance.

[0039] In one example, a real-time drive control signal sequence is calculated based on riding control requirement parameters, and the front and rear wheel motor torques are dynamically allocated to the real-time drive control signal sequence to obtain a target motor drive execution instruction set, including: The riding control demand parameters are input into the fractional-order elastic controller, and the fractional-order control signal is calculated by setting the order parameters of the fractional-order differential operator and the fractional-order integral operator in the fractional-order elastic controller and combining the proportional, integral and differential gain coefficients; Establishing elastic response control instructions between speed control and position control based on fractional-order control signals; The elastic response control instructions are constrained by Lyapunov energy function and time delay correction to obtain a real-time drive control signal sequence; According to the road adhesion conditions and current state of the scooter, the front and rear wheel torques are distributed to the real-time drive control signal sequence to obtain the target motor drive execution instruction set.

[0040] In this example, the riding control demand parameters are input into a fractional-order elastic controller. This controller differs from a traditional integer-order proportional-integral-derivative controller by incorporating a fractional-order differential operator. This allows the controller to retain a stronger memory of historical states in the time domain and a wider regulation bandwidth in the frequency domain, enabling it to respond more flexibly to nonlinear perturbations and high dynamics of the sensory input. To ensure a balance between controller response accuracy and regulation characteristics, the order parameters of the fractional-order differential and integral operators are set based on the system's dynamic response performance. The differential order is set between 0.3 and 0.8 to enhance the system's sensitivity to future trends, while the integral order is controlled between 0.2 and 0.6 to maintain a moderate response to the accumulated error and avoid integral saturation-induced system oscillations. After the operator orders are set, the difference between the sensory input error and the desired behavioral target is weighted and combined with the proportional, integral, and differential gain coefficients to generate a set of structured fractional-order control signals. Based on fractional-order control signals, a set of elastic response control instructions is constructed, which lies between speed control and position control. Rather than using absolute position or speed as a strong constraint, these control instructions establish a flexible target range between the two. This prevents the scooter from experiencing stability issues due to accumulated position errors, nor from sudden jumps due to excessively rigid speed control during dynamic adjustment. By adjusting the control target within a sliding window between speed and position, an acceptable range is established, thereby enhancing the naturalness of the riding experience and the robustness of the controller. To ensure that the elastic response control instructions do not lead to energy overload, motion instability, or system overshoot during execution, a Lyapunov energy function is introduced to evaluate the energy constraints of the controller output. Specifically, the total kinetic and potential energy combination of the scooter-rider system is calculated within each control cycle and compared with a set maximum allowable energy threshold. When the system output control signal is about to cause the total energy to exceed the stability boundary, the control signal is automatically compressed or rotated along the energy gradient, ensuring that the actual control path always lies in the descending direction of the energy function, ensuring full convergence of the entire control behavior. At the same time, due to multiple sources of hysteresis in the scooter control system, such as signal acquisition delay, computational delay, and motor response hysteresis, the elastic control instructions are time-delayed and corrected. By constructing a set of predictive models, using the current state parameters as initial values, the system infers the state change trend over a period of time in the future. Based on this, the current control output is compensated in advance, resulting in a forward-looking real-time drive control signal sequence. This ensures that the controller output and the actual system response are aligned in time, preventing control lag from causing system instability. The real-time drive control signal sequence is then used to distribute front and rear wheel torque based on the scooter's road adhesion conditions and current state.This process combines wheel speed difference, vehicle pitch angle trends, slip rate fluctuations, and a ground adhesion coefficient estimation model to determine the current front and rear wheel adhesion distribution of the scooter. A torque distribution coefficient model is constructed based on the actual vehicle speed and power request. During high-adhesion straight-line acceleration, more drive torque is preferentially allocated to the front wheels to improve traction efficiency. During low-adhesion or cornering situations, the rear wheel torque contribution is appropriately increased to enhance vehicle stability and lateral control. After torque distribution is completed, a torque rate-limiting mechanism is applied to the resulting distribution. A continuous control function calculates the difference between the output of the previous control cycle and the current target output, limiting the change within the system's dynamic range to prevent motor shock or instantaneous tire slip caused by torque jumps. The control signals, after dynamic distribution, smoothing, and stability correction, are encoded into target motor control commands for the front and rear wheels, including specific execution variables such as current modulation parameters, torque request value, and control cycle reference frequency. These signals are then transmitted to the motor drive units, completing the closed-loop path from motion recognition to drive response.

[0041] In one example, based on the road adhesion conditions and current state of the scooter, the front and rear wheel torques are distributed to the real-time drive control signal sequence to obtain the target motor drive execution instruction set, including: Analyze the adhesion characteristics of the scooter's current operating surface to obtain an adhesion characteristic index, and calculate the front and rear wheel torque distribution coefficient based on the adhesion characteristic index combined with the scooter's wheel speed and acceleration data; Decompose the real-time drive control signal sequence based on the front and rear wheel torque distribution coefficients, combine the longitudinal control component and the lateral control component according to the set steering coefficient, and calculate the front wheel torque command and the rear wheel torque command respectively; Applying torque change rate limits to the front wheel torque command and the rear wheel torque command to obtain a torque output value with a smooth transition; The smoothly transitioned torque output value is compared with the current safety state, and the safe control mode is smoothly switched to in an abnormal state to obtain the target motor drive execution instruction set.

[0042] In this example, real-time adhesion analysis of the scooter's current road surface is performed. This process relies on data collected by accelerometers, wheel speed sensors, and gyroscopes to comprehensively assess the impact of factors such as surface material, surface moisture, fine particle content, and local slope on tire-to-ground friction. The system detects signs of wheel slip by identifying the nonlinear hysteresis relationship between wheel slip and vehicle acceleration response. Based on this information, the system estimates the maximum adhesion of each front and rear wheel. Furthermore, it considers the additional effects of steering adhesion symmetry and posture coupling between the left and right wheels to construct a set of adhesion characteristics indicators representing the adhesion of the front and rear wheels. These indicators reflect the traction limit of the wheels under current operating conditions and indicate whether the vehicle's motion is in a stable mechanical region. The adhesion characteristics indicators are integrated with wheel speed and longitudinal acceleration data to generate a set of front and rear wheel torque distribution coefficients for dynamic torque resource scheduling. During vehicle acceleration, if the front wheels detect high adhesion and the rear wheels approach their limits, more drive force is allocated to the front wheels. Conversely, when vehicle stability or steering response needs to be improved, the rear wheels' torque share is appropriately increased to generate a stronger rear-end correction torque. Based on this allocation coefficient, the current real-time drive control signal sequence is effectively decomposed, separating the vehicle's longitudinal control commands from its lateral posture adjustment commands. The output commands for the front and rear wheels are then synthesized according to the set steering coefficient. The steering coefficient is set based on the scooter's geometric parameters, the vehicle's center of gravity, the rider's posture distribution, and the distance between the front and rear wheels. It represents the relationship between the scooter's turning radius and the inter-wheel coupling torque, enabling precise mapping of the unified control signal into independent target torque commands for each drive motor. After calculating the front and rear wheel torque commands, these preliminary torque commands are subject to rate limits, taking into account the dynamic response limits and structural load constraints of the drive motors, to prevent sudden changes in the commands from causing structural shock, current spikes, or sudden increases in tire loads. In its implementation, the current torque command is compared with the output value from the previous control cycle, the instantaneous change is calculated, and then compared with the set maximum allowable rate of change. If the change exceeds the upper limit, it is clipped and smoothed according to the maximum change amplitude, thereby ensuring continuous and gradual torque output for the front and rear wheels during command execution. This mechanism helps ensure stable closure of the motor drive control loop and effectively reduces peak energy loss, extending the life of the battery and drive components. It also enhances the human perception of vehicle response and prevents abrupt pushes and pulls that can cause imbalance or psychological discomfort for the rider. After smoothing, the current torque output value is compared and analyzed with the scooter's real-time operational safety status. This process relies on a dynamic safety state assessment module, which constructs a set of state functions based on multi-dimensional variable criteria, covering multiple risk indicators including lateral stability, longitudinal acceleration overshoot, attitude angle deviation, speed mutation, grade critical point, and tire slip threshold.When the system determines that any indicator is approaching a preset safety threshold, it triggers the control mode's safety switching mechanism, entering a conservative control zone prioritized for safety. In this mode, the system no longer operates entirely according to the original control logic. Instead, it performs a weighted fusion of the current control output between conventional control and safety control. This allows the control signal to gradually reduce its output amplitude, compress its control range, or slowly adjust its output trajectory while maintaining its original directionality and response trend, to prevent the vehicle from entering a zone of mechanical instability. This fusion process is constructed using a continuous smooth transition function to ensure that the switching process does not cause sudden jumps or signal discontinuities, allowing the scooter to maintain a balanced state in complex dynamic scenarios, minimizing serious consequences such as posture collapse, tire slippage, and drive loss of control. The smooth transition torque output value after the above decomposition, constraint and safety switching processing is converted into a target motor drive execution instruction set that conforms to the driver control protocol. This instruction set contains control variables such as the target output torque of the front and rear wheels, current regulation target, motor response curve coefficient, voltage control mode switching parameters, maximum power protection threshold, etc., and is then transmitted to the motor controller through the drive bus at a specified control frequency. The motor controller adjusts the PWM duty cycle, current sampling feedback and position encoder output in real time according to the target instruction, realizing the scooter's precise coordinated response to longitudinal traction, lateral stability and energy consumption control under the current somatosensory input background.

[0043] Reference Figure 2 This embodiment provides a somatosensory drive control system for a scooter, including: Acquisition module 1, used to collect the original body sensory signals of the scooter; Amplification module 2, used to perform resonant amplification and multi-modal mixing on the original body sensory signal to obtain a body sensory characteristic signal; State calculation module 3, used to establish a scooter-rider coupling system dynamics model based on the somatosensory characteristic signals, and perform real-time state calculation to obtain current state parameters and state deviation values; Mapping conversion module 4, used to perform somatosensory weight distribution change mapping conversion on the current state parameter and the state deviation value to obtain the riding control demand parameter; The dynamic allocation module 5 is used to calculate the real-time drive control signal sequence according to the riding control requirement parameters, and dynamically allocate the front and rear wheel motor torques of the real-time drive control signal sequence to obtain the target motor drive execution instruction set.

[0044] In this embodiment, for the specific implementation of each unit in the above system embodiment, please refer to the above method embodiment, which will not be repeated here.

[0045] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, system, article, or method comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, system, article, or method. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, system, article, or method comprising the element.

[0046] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A somatosensory drive control method for a scooter, characterized in that: include: Collect the original somatosensory signals of the scooter; Performing resonance amplification and multi-modal mixing on the original somatosensory signal to obtain a somatosensory characteristic signal; Establishing a scooter-rider coupling system dynamics model based on the somatosensory characteristic signal, and performing real-time state calculation to obtain current state parameters and state deviation values; Performing a somatosensory weight distribution change mapping conversion on the current state parameter and the state deviation value to obtain a riding control demand parameter; A real-time drive control signal sequence is calculated according to the riding control demand parameter, and the front and rear wheel motor torques are dynamically distributed to the real-time drive control signal sequence to obtain a target motor drive execution instruction set.

2. The somatosensory driving control method of a scooter according to claim 1, characterized in that: The collecting of the original somatosensory signal of the scooter includes: A micro-accelerometer array is installed on the pedal surface of the scooter, pressure sensors are set on both sides of the pedal, and a six-axis inertial measurement unit is assembled at the central control unit to obtain a somatosensory data acquisition network; The body sensing data acquisition network is used to acquire the rider's weight distribution change data and the scooter's posture data in real time to obtain the original sensing signal.

3. The somatosensory driving control method of a scooter according to claim 1, characterized in that: The performing resonant amplification and multi-modal mixing on the original body sensory signal to obtain a body sensory characteristic signal includes: Inputting the original body sensory signal into an LC resonant circuit array, and adjusting the inductance and capacitance parameters of the LC resonant circuit array so that the resonant frequency matches the characteristic frequency range of human motion, thereby obtaining a resonantly amplified body sensory signal; Performing a combined operation on the resonantly amplified body sensory signals through a multimodal physical mixing circuit to obtain a comprehensive characteristic signal of weight distribution changes; Inputting the comprehensive characteristic signal of weight distribution change into a phase-locked amplifier circuit, calculating a phase ratio between a reference signal in the phase-locked amplifier circuit and the comprehensive characteristic signal of weight distribution change, and amplifying a valid signal in the comprehensive characteristic signal of weight distribution change according to the phase ratio, thereby obtaining a body sensing signal with a high signal-to-noise ratio; The direction and acceleration information of the weight distribution change are extracted according to the high signal-to-noise ratio somatosensory signal to obtain a somatosensory characteristic signal.

4. The somatosensory driving control method of a scooter according to claim 1, characterized in that: The method of establishing a scooter-rider coupling system dynamics model based on the somatosensory characteristic signal and performing real-time state calculation to obtain current state parameters and state deviation values includes: Extracting the position coordinates, center of gravity distribution, and posture data of the rider on the pedal based on the somatosensory feature signals, and combining them with the scooter body parameters to obtain a state observation dataset; Based on the state observation data set, a system configuration description matrix of the scooter and the rider is constructed and a scooter-rider coupled system dynamics model including a mass matrix, a Coriolis force matrix and a gravity term is derived; Constructing a mirror state estimator based on the scooter-rider coupling system dynamics model, inputting the somatosensory characteristic signal and inertial measurement unit data into the mirror state estimator, performing state prediction and measurement update using an extended Kalman filter algorithm to obtain current state parameters; A state deviation value between the ideal state and the current state is calculated based on the current state parameters and a preset ideal riding model.

5. The somatosensory driving control method for a scooter according to claim 4, characterized in that: The method comprises: constructing a system configuration description matrix of the scooter and the rider based on the state observation data set and deriving a scooter-rider coupled system dynamics model including a mass matrix, a Coriolis force matrix, and a gravity term. Combining the center of mass position, orientation angle, and rider posture angle of the scooter into a system configuration description matrix based on the state observation data set; Calculating mass attributes of the system configuration description matrix, analyzing mass distribution and inertia parameters of various parts of the scooter and the rider, and obtaining a system mass matrix; Calculating velocity coupling characteristics based on the system configuration description matrix and the system mass matrix, and constructing a Coriolis force matrix through operations on velocity derivatives and angular velocity; A gravitational potential energy analysis is performed on the system configuration description matrix to calculate the gravity term, and the system mass matrix, the Coriolis force matrix and the gravity term are combined to construct a scooter-rider coupling system dynamics model.

6. The somatosensory driving control method for a scooter according to claim 4, characterized in that: The mirror state estimator is constructed according to the dynamic model of the scooter-rider coupling system, the somatosensory characteristic signal and the inertial measurement unit data are input into the mirror state estimator, and the state prediction and measurement update are performed through the extended Kalman filter algorithm to obtain the current state parameters, including: A mirror state estimator is constructed based on the dynamic model of the scooter-rider coupling system, the system state variables are organized into a state vector containing position, angle and its derivatives, and a state transfer function and a measurement function are set to obtain a state space description; Performing time update processing on the state vector in the state space description, and calculating the state prediction value and the prediction covariance matrix according to the system dynamics equation to obtain preliminary predicted state data; Using the somatosensory characteristic signal and the inertial measurement unit data as observation inputs, calculating the deviation between the preliminary predicted state data and the actual measurement value, and calculating the Kalman gain coefficient based on the measurement noise matrix and the prediction covariance matrix to obtain a state correction factor; The state correction factor is applied to correct the preliminary predicted state data, and the system state covariance matrix is updated to obtain the current state parameters.

7. The somatosensory driving control method for a scooter according to claim 1, characterized in that: The performing a somatosensory weight distribution change mapping conversion on the current state parameter and the state deviation value to obtain a riding control demand parameter includes: Analyzing a change pattern of the rider's weight distribution according to the current state parameters, determining a center of gravity offset and a change rate in the front, rear, left, and right directions based on the change pattern, and constructing a weight distribution change vector according to the center of gravity offset and the change rate; Performing error function calculation on the weight distribution change vector and the state deviation value to obtain a control deviation matrix; Deconstructing the control deviation matrix and calculating the control law, and converting the control deviation matrix into the required driving force and steering force to obtain basic driving demand data; According to the basic driving demand data and the current operating state of the scooter, the Jacobian matrix is calculated and combined with the proportional-differential controller to generate the riding control demand parameters.

8. The somatosensory driving control method for a scooter according to claim 1, characterized in that: The step of calculating a real-time drive control signal sequence according to the riding control requirement parameters, and dynamically allocating the front and rear wheel motor torques to the real-time drive control signal sequence to obtain a target motor drive execution instruction set includes: Inputting the riding control demand parameter into a fractional-order elastic controller, and calculating a fractional-order control signal by setting order parameters of a fractional-order differential operator and a fractional-order integral operator in the fractional-order elastic controller and combining proportional, integral, and differential gain coefficients; Establishing an elastic response control instruction between speed control and position control based on the fractional-order control signal; Performing Lyapunov energy function constraints and time delay correction on the elastic response control instructions to obtain a real-time drive control signal sequence; According to the road adhesion condition and current state of the scooter, the front and rear wheel torques are distributed to the real-time drive control signal sequence to obtain a target motor drive execution instruction set.

9. The somatosensory driving control method for a scooter according to claim 8, characterized in that: The method of allocating front and rear wheel torques to the real-time drive control signal sequence based on the road adhesion condition and current state of the scooter to obtain a target motor drive execution instruction set includes: Analyzing the adhesion characteristics of the current road surface on which the scooter is operating to obtain an adhesion characteristic index, and calculating the front and rear wheel torque distribution coefficient based on the adhesion characteristic index combined with the scooter's wheel speed and acceleration data; Decomposing the real-time drive control signal sequence based on the front and rear wheel torque distribution coefficients, combining the longitudinal control component and the lateral control component according to the set steering coefficient, and respectively calculating the front wheel torque command and the rear wheel torque command; Applying a torque change rate limit to the front wheel torque command and the rear wheel torque command to obtain a torque output value with a smooth transition; The smoothly transitioned torque output value is compared with the current safety state, and the safe control mode is smoothly switched to under abnormal conditions to obtain a target motor drive execution instruction set.

10. A somatosensory drive control system for a scooter, characterized in that: The steps for implementing the somatosensory drive control method of the scooter according to any one of claims 1 to 9, wherein the somatosensory drive control system of the scooter comprises: The acquisition module is used to collect the original body sensory signals of the scooter; an amplification module, configured to perform resonance amplification and multi-modal mixing on the original somatosensory signal to obtain a somatosensory characteristic signal; A state calculation module is used to establish a scooter-rider coupling system dynamics model based on the somatosensory characteristic signal and perform real-time state calculation to obtain current state parameters and state deviation values; a mapping conversion module, configured to perform a somatosensory weight distribution change mapping conversion on the current state parameter and the state deviation value to obtain a riding control demand parameter; The dynamic allocation module is used to calculate a real-time drive control signal sequence according to the riding control requirement parameters, and dynamically allocate the front and rear wheel motor torques of the real-time drive control signal sequence to obtain a target motor drive execution instruction set.

Images