Unmanned vehicle control algorithm with emergency obstacle avoidance function

Abstract

The invention relates to the technical field of unmanned vehicle control, and discloses an unmanned vehicle control algorithm with an emergency obstacle avoidance function, which comprises the following steps: acquiring a transverse deviation between a current attitude of a vehicle and an obstacle avoidance target trajectory, and determining an error flux according to a product of a first derivative of the transverse deviation and a curvature change rate of the obstacle avoidance target trajectory; converting the error flux into a gain adjustment factor based on a preset nonlinear mapping function, and performing real-time weighted weakening on a proportional gain and a differential gain in a steering closed-loop control loop, so that a steering actuator is in a controlled under-gain state in an obstacle avoidance cut-in stage; after the error flux falls back to the physical linear threshold value, feedback gain is recovered according to the energy recovery curve, physical matching of the control rigidity and input sudden change is achieved, the nonlinear saturation risk of an actuator is avoided, and the loop stability of a control system under the extreme working condition is guaranteed.

Description

Technical Field

[0001] This invention belongs to the field of autonomous vehicle control technology, and in particular relates to an autonomous vehicle control algorithm with emergency obstacle avoidance function. Background Technology

[0002] Current autonomous vehicle control systems utilize feedback regulators to maintain the convergence of state variables to the target trajectory, while conventional operating condition controllers use fixed control gains to eliminate trajectory deviations and drive chassis actuators to perform lateral and longitudinal movements.

[0003] In emergency obstacle avoidance scenarios, the rate of curvature change of the avoidance path generated by the obstacle avoidance algorithm often exceeds the physical response limit of the vehicle chassis. The feedback controller treats the step deviation as a normal error and attempts to eliminate the deviation using a fixed control stiffness. The energy of the feedback loop increases sharply due to the sudden change in input, causing the steering actuator to produce saturation and desaturation nonlinear oscillations, resulting in mechanical wear of chassis actuators and vehicle attitude instability. For example, Chinese invention patent CN114559923B discloses an automatic emergency obstacle avoidance system and its control method for unmanned vehicles. It introduces the consensus behavior transfer algorithm CBMA combined with circular clustering for obstacle avoidance trajectory planning. Its center of gravity is at the static planning level of spatial trajectory points. The path is selected by iteratively evaluating the fitness function. The high-speed obstacle avoidance scenario introduces computational delay. Such algorithms do not address the deep harmonization between control strength and physical input mutation, and lack an active physical weakening mechanism for feedback gain. The actuator faces the risk of nonlinear saturation when facing large transient deviations. After obstacle avoidance, trajectory fluctuations or convergence phase differences occur, making it difficult to achieve dynamic alignment between control stiffness and the real-time load-bearing capacity of the hardware.

[0004] Therefore, how to dynamically adjust the feedback gain based on the error flux to avoid the risk of actuator saturation, and achieve smooth convergence of the obstacle avoidance process while reducing the processor's computing power burden, has become the technical problem to be solved by this invention. Summary of the Invention

[0005] This invention provides a control algorithm for an unmanned vehicle with emergency obstacle avoidance function, comprising the following steps:

[0006] Step S101: Real-time acquisition of the lateral deviation between the vehicle's current lateral attitude and the obstacle avoidance target trajectory, and determination of the error flux based on the product of the first derivative of the lateral deviation and the rate of change of curvature of the obstacle avoidance target trajectory.

[0007] Step S102: Input the error flux into a preset nonlinear mapping function to determine the gain adjustment factor; wherein, the nonlinear mapping function limits the gain adjustment factor to have an inverse proportional mapping relationship with the absolute value of the error flux;

[0008] Step S103: The proportional gain and derivative gain in the steering closed-loop control loop are weighted and weakened in real time using a gain adjustment factor, so that the rate of change of the steering actuator's action command is within the rated physical response envelope of the steering actuator, and the vehicle is in a controlled undergain state.

[0009] Step S104: Continuously monitor the change in error flux, and determine the preset energy recovery curve when the error flux is less than or equal to the preset physical linearity threshold.

[0010] Step S105: The proportional gain and differential gain are restored cycle by cycle according to the slope of the energy recovery curve, and the adjusted control quantity is output to drive the steering actuator to perform obstacle avoidance action.

[0011] Preferably, step S102 specifically includes: calculating the product of the first derivative of the lateral deviation and the rate of change of curvature of the obstacle avoidance target trajectory in real time to determine the adjustment pressure value; mapping the adjustment pressure value to a gain adjustment factor using a preset ramp function; wherein, the gain adjustment factor monotonically decreases as the absolute value of the adjustment pressure value increases; the value range of the gain adjustment factor is limited to between 0.2 and 0.8.

[0012] Preferably, it further includes: acquiring the slip ratio parameter of the vehicle wheel end in real time to determine the road surface adhesion coefficient; dynamically constraining the change slope of the gain adjustment factor based on the road surface adhesion coefficient; wherein, when the road surface adhesion coefficient is less than a preset adhesion threshold, the attenuation rate of the proportional gain and the derivative gain is reduced.

[0013] Preferably, step S103 further includes: determining the safe side orientation of the obstacle avoidance target trajectory relative to the obstacle; decomposing the lateral deviation into the local coordinate system corresponding to the safe side orientation to obtain the safe side lateral component and the dangerous side lateral component; and using a preset weighting coefficient for the safe side lateral component and a normal gain coefficient greater than the weighting coefficient for the dangerous side lateral component to perform the feedback gain reduction action in the direction away from the obstacle.

[0014] Preferably, during the execution of step S103, the method further includes: superimposing a micro-disturbance signal of a preset frequency onto the target control command of the steering actuator; acquiring the position feedback signal of the steering actuator and extracting the response features in the position feedback signal that are at the same frequency as the micro-disturbance signal; and identifying the dynamic impedance of the steering actuator based on the response features. Among them, dynamic impedance Satisfying the formula: ,in, The amplitude of the micro-perturbation signal superimposed on the target control command is denoted as . The response amplitude of the position feedback signal. The phase difference between the position feedback signal and the micro-perturbation signal; utilizing dynamic impedance Adjust the recovery step size of the energy recovery curve.

[0015] Preferably, the method further includes: calculating the deviation residual of the feedback gain before and after adjustment; mapping the deviation residual to a longitudinal load adjustment signal; adjusting the braking distribution ratio of the front axle and rear axle of the vehicle based on the longitudinal load adjustment signal to increase the vertical load on the vehicle's steering wheels.

[0016] Preferably, step S105 specifically includes: monitoring the curvature direction switching point of the obstacle avoidance target trajectory; when reaching the curvature direction switching point, determining a phase compensation pulse based on the current lateral deviation value; and injecting the phase compensation pulse into the steering closed-loop control loop to counteract the vehicle's accumulated lateral kinetic energy.

[0017] Preferably, step S101 further includes: smoothing the lateral deviation using a low-pass filter to filter out noise components higher than a preset cutoff frequency; the preset cutoff frequency is half of the steering actuator sampling frequency.

[0018] Preferably, after step S105 is executed, the method further includes: extracting the drive current fluctuation of the steering actuator in real time; pausing the recovery process corresponding to the energy recovery curve when the drive current fluctuation is greater than a preset current threshold; maintaining the proportional gain and differential gain at the current level until the drive current fluctuation is less than or equal to the preset current threshold.

[0019] Preferably, the energy recovery curve satisfies the second-order critical damping characteristics; the recovery period of the energy recovery curve is 1.5 to 2.5 times the time required for the steering actuator to complete the rated stroke.

[0020] Compared with existing technologies, the autonomous vehicle control algorithm of this invention with emergency obstacle avoidance function has the following advantages:

[0021] 1. In the control algorithm of autonomous vehicles, the transient weakening of feedback gain is achieved by adjusting the error flux, changing the inherent pattern of increasing correction intensity as the error increases. Short-term trajectory deviation is actively allowed at the moment the obstacle avoidance command is engaged, avoiding the risk of nonlinear saturation of actuators and ensuring the stability of the control system loop under extreme operating conditions. By microscopically deconstructing the error flux, the physical matching between control stiffness and input mutation is achieved, reducing the peak pulse pressure on the steering mechanism during obstacle avoidance transients and extending the service life of chassis actuators.

[0022] 2. By using wheel-end slip ratio to characterize the road surface adhesion level and dynamically constrain the rate of gain reduction, the control stiffness change process is aligned with the ground physical limit logic, maintaining the upper limit of lateral guidance force in low adhesion environment of the vehicle, avoiding the risk of runaway caused by the mismatch between pure logic stiffness relaxation and underlying physical contact conditions, and reusing existing data from the chassis bus to adjust the control strategy, so that the obstacle avoidance action maintains dynamic stability in different friction environments.

[0023] 3. Based on the identification results of the target trajectory relative to the safe lateral orientation of the obstacle, the asymmetric weight projection of the adjustment error component is used to guide the displacement margin generated by the gain reduction to a spatial region away from the obstacle, thereby achieving the direction of the deterministic deviation of the flexible control state. Without increasing the output strength of the actuator, the spatial hard constraint requirements of the obstacle avoidance process are met. By asymmetrically modifying the error weight distribution, the loss-of-track deviation is converted into obstacle avoidance gain, so that the flexible control logic and the spatial collision avoidance boundary can work together. Attached Figure Description

[0024] Fig. 1 This is a schematic diagram of the hardware topology and logic architecture of the unmanned vehicle control system of the present invention;

[0025] Fig. 2 This is a comparison curve of the vehicle lateral deviation convergence performance under different control strategies of the present invention;

[0026] Fig. 3 This is a flowchart of the emergency obstacle avoidance control algorithm based on error flux perception of the present invention. Detailed Implementation

[0027] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0028] It should be noted that all directional and positional terms used in this invention, such as: up, down, left, right, front, back, vertical, horizontal, inner, outer, top, low, lateral, longitudinal, center, etc., are only used to explain the relative positional relationship and connection between components in a specific state (as shown in the accompanying drawings). They are only for the convenience of describing this invention and do not require that this invention be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting this invention. In addition, the descriptions of "first," "second," etc., in this invention are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly indicating the number of technical features indicated.

[0029] In the description of this invention, unless otherwise explicitly specified and limited, the terms installation, connection, and linking should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; they can refer to mechanical connections; they can refer to direct connections or indirect connections through an intermediate medium; they can refer to the internal connection of two components. For those skilled in the art, the specific meaning of the above terms in this invention can be understood according to the specific circumstances.

[0030] In the description of this specification, references to the terms "an embodiment," "some embodiments," "illustrative embodiments," "examples," "specific examples," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example, and the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0031] The present invention will be described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the following embodiments are intended to explain the present invention and not to limit the scope of protection of the present invention.

[0032] An autonomous vehicle control algorithm with emergency obstacle avoidance capability comprises four stages: error flux extraction, gain adjustment factor mapping, feedback gain transient attenuation, and energy smoothing recovery. The system monitors the dynamic error characteristics between the vehicle's lateral attitude and the obstacle avoidance target trajectory, and uses a nonlinear mapping function to adjust the control strength of the steering closed-loop control loop in real time. This ensures the steering actuator is in a controlled under-gain state during the obstacle avoidance engagement phase, thus avoiding the risk of actuator nonlinear saturation due to step-like deviation input and guaranteeing the loop stability of the control system under operating conditions. To address the sudden increase in feedback loop energy caused by abrupt changes in path curvature during emergency obstacle avoidance, the system executes an error flux feature extraction procedure, using an onboard inertial measurement unit and positioning system to obtain the vehicle's current lateral attitude and determine the current lateral deviation based on the obstacle avoidance target trajectory. The first derivative of the lateral deviation is obtained using a first-order difference element. Simultaneously extract the rate of curvature change of the obstacle avoidance target trajectory. The system will use the first derivative of the lateral deviation. Rate of curvature change of the obstacle avoidance target trajectory The product of these factors is determined as the error flux. This value objectively characterizes the transient adjustment pressure borne by the feedback loop. During execution, if the vehicle... Speed ​​and path curvature A sudden change in the rate of change, when the resulting lateral deviation changes at a rate of... At that time, the calculated error flux for This value provides a deterministic logical input for subsequent control stiffness adjustment.

[0033] Since a fixed control gain is prone to inducing saturation oscillations in the steering actuator when facing a step deviation, the system employs a nonlinear gain mapping procedure to reduce the error flux. Input a preset nonlinear mapping function to determine the gain adjustment factor. The mapping function limits the gain adjustment factor. With error flux The absolute values ​​exhibit an inverse proportional mapping relationship. In this embodiment, the system uses a ramp function to perform the mapping and adjusts the gain factor. The range of values ​​is constrained by to Between, using gain adjustment factor Proportional gain in the steering closed-loop control loop and differential gain Perform real-time weighted weakening by calculation as well as The adjusted feedback gain is determined so that the rate of change of the steering actuator's action command is within the rated physical response envelope of the steering actuator. The controlled undergain state actively allows for short-term trajectory deviations, achieving a physical match between control strength and input abrupt changes; physical linear threshold. Acquisition method: A slope error simulation signal is injected into the steering control loop by a static test bench, and the drive current of the steering actuator is collected simultaneously. With output torque Monitoring the ratio between the two and ensuring it deviates from the linear region by more than [a certain amount]. Critical inflection point, and its corresponding error flux numerical values Defined as The construction of nonlinear mapping functions includes Discrete feature point numerical lookup table, error flux Absolute value exceeds When using formula Calculate the gain adjustment factor The limit is in to Between, determine the proportional gain With differential gain Transient attenuation of strength.

[0034] To prevent the vehicle from developing a convergence phase difference due to excessively rapid release of control energy after obstacle avoidance, the system executes energy recovery and state convergence procedures, and continuously monitors the error flux. The variation in error flux, and in error flux Less than or equal to the preset physical linear threshold At that time, the energy recovery curve satisfying the second-order critical damping characteristic is determined, and the system adjusts the proportional gain according to the slope of the energy recovery curve. and differential gain A step-by-step recovery is performed, with each control cycle corresponding to a recovery cycle that satisfies the time required for the steering actuator to complete its rated stroke. Doubled The system incorporates a relaxation slope correction mechanism guided by physical adhesion margin, taking into account the impact of different road surface adhesion conditions on control performance. The system also collects the slip ratio parameters at the vehicle wheel ends in real time. To determine the road surface adhesion coefficient And based on the road surface adhesion coefficient Gain adjustment factor Dynamic constraints are applied to the changing slope of the road surface adhesion coefficient. When the gain is less than the preset attachment threshold, the system reduces the proportional gain. and differential gain The system reduces the rate of deviation; to address the issue that trajectory deviations may encounter hard spatial constraints in complex environments, the system executes an error reprojection procedure tilted towards the obstacle avoidance safety side, determines the safe side orientation of the obstacle avoidance target trajectory relative to the obstacle, and incorporates the lateral deviation. Decomposed to the local coordinate system corresponding to the safe side orientation, the safe side lateral component is obtained. and the lateral component on the dangerous side The system is based on the security-side lateral component. With preset weighting coefficients And based on the lateral component of the danger side By using a normal gain coefficient greater than the weighting coefficient, the weakening action of the feedback gain is preferentially executed in the direction away from the obstacle. This asymmetric error weight distribution induces the displacement margin generated by the vehicle to be oriented to the safe area, thus meeting the spatial obstacle avoidance requirements without increasing the output strength of the actuator.

[0035] To address the dynamic impedance drift of the steering actuator under load, the system employs a command-stream micro-excitation echo analysis mechanism. During the transient attenuation of the feedback gain, a micro-perturbation signal of a preset frequency is superimposed onto the target control command of the steering actuator. The position feedback signal of the steering actuator is acquired, and response characteristics with the same frequency as the micro-perturbation signal are extracted. Based on these response characteristics, the dynamic impedance of the steering actuator is identified. Dynamic impedance Satisfying the formula: ,in, The amplitude of the micro-perturbation signal superimposed on the target control command. The response amplitude of the position feedback signal. The phase difference between the position feedback signal and the micro-perturbation signal is calculated using dynamic impedance. The recovery step size of the energy recovery curve is adjusted to dynamically align the control stiffness recovery process with the actual hardware response capability, eliminating phase deviation caused by actuator overload hysteresis. For the residual component generated in lateral control adjustment, a load recirculation adjustment procedure is designed. The residual deviation of the feedback gain before and after adjustment is calculated and mapped to a longitudinal load adjustment signal. Based on this signal, the braking distribution ratio of the front and rear axles is adjusted, driving the vehicle's center of gravity forward to increase the vertical load on the steering wheels. This procedure compensates for the lateral obstacle avoidance control energy in the longitudinal stability by increasing tire lateral stiffness. The longitudinal load adjustment signal is generated based on the residual deviation before and after feedback gain adjustment, and the proportional gain is calculated. The reduction is converted into an electronic braking system pressure distribution bias, keeping the total braking torque constant, and the front axle to rear axle braking pressure ratio changes from the initial... Dynamically adjusted to the highest level Inducing the vehicle's center of gravity to shift forward increases the vertical load on the steering wheels. N to N compensates for lateral stiffness; Proportional gain recovery step size within ms sampling period Based on real-time identification of dynamic impedance correction, Superimposed by target instructions Hz sinusoidal micro-perturbation signal amplitude Position feedback signal response amplitude and phase difference The calculation shows that the rate of increase in control stiffness is matched with the real-time load characteristics of the steering mechanism.

[0036] At the curvature switching point of the obstacle avoidance trajectory, the system executes a phase lag pre-compensation procedure, monitors the curvature direction switching point of the obstacle avoidance target trajectory, and upon reaching the curvature direction switching point, adjusts the current lateral deviation accordingly. The phase compensation pulse is numerically determined and injected into the steering closed-loop control loop to counteract the lateral kinetic energy accumulated by the vehicle during the feedback gain attenuation phase, thus resolving the insufficient self-centering issue after emergency avoidance. To ensure the quality of the control signal, the system utilizes a low-pass filter to filter lateral deviations. The system performs smoothing processing to filter out noise components higher than a preset cutoff frequency. The preset cutoff frequency is half of the steering actuator sampling frequency. Simultaneously, the drive current fluctuation of the steering actuator is extracted in real time. When the drive current fluctuation exceeds a preset current threshold, the recovery process corresponding to the energy recovery curve is paused, and the feedback gain is maintained at the current level until the drive current fluctuation is less than or equal to the preset current threshold, thereby ensuring the reliability of the actuator under complex loads. The method and its steps claimed in this invention can run in vehicle controllers such as Qualcomm 8295 or NVIDIA Orin. The system achieves stable control of the obstacle avoidance process by dynamically deconstructing the data flow inside the steering control loop.

[0037] Example 1: When an unmanned vehicle... Speed ​​and coefficient of friction for When the surface is detected, the front is identified. Obstacles and planning the rate of curvature change for The system obtains the lateral deviation of the emergency avoidance trajectory. first derivative for Based on the first derivative of the lateral deviation Rate of curvature change of the obstacle avoidance target trajectory The product determines the error flux. for The system determines the gain adjustment factor based on the nonlinear mapping function. for Using gain adjustment factor Proportional gain in the steering closed-loop control loop and differential gain Reduced to the rated value This allows the steering actuator to be in a controlled undergain state during the obstacle avoidance engagement phase, thus allowing for short-term trajectory deviation and preventing current saturation and nonlinear oscillations caused by the steering actuator attempting to eliminate step deviations.

[0038] While performing gain reduction, the system maps the reduced feedback energy residual as a longitudinal load adjustment signal, and adjusts the braking pressure distribution ratio between the front and rear axles through the electronic braking system, driving the vehicle's center of gravity forward to increase the vertical load on the front axle. By utilizing the physical characteristic that tire lateral stiffness increases with vertical load, the loss of control torque caused by the decrease in gain is compensated, thereby achieving synergistic enhancement of lateral control logic and longitudinal dynamic load distribution; when the error flux... As the trajectory converges to The following section describes how the system adjusts the proportional gain based on the energy recovery curve that satisfies the second-order critical damping characteristic. and differential gain Perform step-by-step recovery, in During the recovery period, the control intensity was restored to the rated level, and the vehicle ultimately achieved a lateral deviation of less than [missing information]. The state achieves attitude convergence.

[0039] Example 2: In a verification experiment conducted using a hardware-in-the-loop simulation platform, i.e., an experimental environment consisting of a high-performance real-time processor and input / output interfaces, the system evaluated a control algorithm for an unmanned vehicle with emergency obstacle avoidance capabilities. The experimental data originated from a real-time simulation system integrating a seven-degree-of-freedom nonlinear vehicle dynamics model, and was actively injected via the chassis bus with a signal-to-noise ratio of [missing information]. Gaussian white noise is used to simulate electromagnetic interference in industrial environments. The system adjusts the sampling period. Perform calibration to identify factors affecting the sampling period. The main factors influencing the value are the vehicle's sideslip angle dynamic characteristic frequency and the processor instruction cycle load, among which the sampling period... The settings aim to balance the real-time performance of data acquisition with the computational redundancy of the processor. As the spectral bandwidth of the monitored lateral deviation signal increases, the sampling period is adjusted to satisfy the Nyquist sampling theorem and avoid signal aliasing. The value tends towards the lower limit of its range, and in this test, the vehicle speed is... Under the operating conditions described above, the sampling period is determined by applying the aforementioned decision rules. Determined as This value ensures the stability of the calculated load while guaranteeing the accuracy of the control loop response.

[0040] To verify the synergistic efficacy of the transient diastolic regulation mechanism of control intensity, a multi-dimensional control system was established, including an experimental group, a control group, a partially missing control group, and an out-of-range control group. The experimental group employed a complete algorithm with error flux extraction, transient gain attenuation, and load recirculation adjustment. The control group used a conventional fixed-gain control strategy. The partially missing control group disabled the longitudinal load compensation stage during gain attenuation, while the out-of-range control group adjusted the gain adjustment factor. The lower limit was adjusted to To detect the performance inflection point, the test was set at a vehicle speed of... Road surface adhesion coefficient for and set the width to In obstacle intrusion scenarios, the system monitors the drive current of the steering actuator. and vehicle lateral deviation To characterize the physical adaptability of the system; the first derivative of the lateral deviation extracted by the system. Rate of change of curvature of obstacle avoidance target trajectory The product determines the error flux. Peak value Table 1 presents the obstacle avoidance performance test data under different control strategies.

[0041] Table 1: Comparison of Experimental Data for Obstacle Avoidance Algorithms

[0042]

[0043] in, The error flux characterizes the transient regulation pressure experienced by the feedback loop. This is the gain adjustment factor, which satisfies the gain adjustment factor. With error flux The absolute values ​​show an inverse proportional mapping relationship; The lateral deviation of the autonomous vehicle's center of gravity relative to the obstacle avoidance target trajectory is expressed in units of 1. ; lateral deviation convergence to The time required within, in units of The technical effect exhibited by the experimental group stems from the synergistic effect between internal features, when the gain adjustment factor... Determined as This leads to proportional gain and differential gain During transient attenuation, the system synchronously maps the resulting deviation residuals into a longitudinal load adjustment signal, driving the vehicle's center of gravity forward and increasing the vertical load on the steering wheels. Utilizing the physical property that tire lateral stiffness increases with vertical load, the test group compensated for the loss of lateral control torque caused by the decrease in gain, so that under the same level of gain reduction, the peak lateral deviation of the test group was reduced. Compared to the partial deletion control group, it decreased This coupling mechanism of lateral flexibility control and longitudinal load enhancement resolves the conflict between adjusting stiffness relaxation and maintaining lateral guidance force in low-adhesion environments, achieving dynamic balance throughout the obstacle avoidance process.

[0044] Example 3: This example combines Figs. 1 to 3 This document describes a control algorithm for an autonomous vehicle with emergency obstacle avoidance capabilities, such as... Fig. 1As shown, the algorithm runs on a high-performance onboard computing platform, connecting the perception and signal input nodes and the chassis actuator nodes via a high-speed communication bus on the vehicle chassis, thus constructing a complete hardware topology. The perception and signal input nodes include an inertial measurement unit that provides vehicle lateral attitude data, a positioning system that provides obstacle avoidance target trajectory information, and a wheel speed sensor group responsible for providing wheel-end slip ratio parameters. The internal logic architecture of the computing platform integrates an error flux calculation module for calculating the product of the deviation derivative and the rate of change of curvature, a nonlinear gain mapping module for generating gain adjustment factors, a steering closed-loop control module responsible for performing real-time weighted attenuation and energy recovery operations, and a dynamic impedance monitoring module responsible for injecting micro-disturbance signals to identify actuator impedance. The chassis actuator nodes include a steering actuator that performs obstacle avoidance steering actions and provides feedback on position, current, and impedance characteristics, as well as an electronic braking system responsible for front and rear axle brake distribution and longitudinal load adjustment.

[0045] like Fig. 2 As shown, in a two-dimensional coordinate system with time as the horizontal axis and lateral deviation e as the vertical axis, the obstacle avoidance performance comparison curves under different control strategies are displayed. The solid line trajectory of the experimental group of the present invention shows the lowest peak lateral deviation e and the fastest convergence time, indicating that the system is in a stable linear response state. The short dashed line trajectory of the control group with fixed gain shows a high peak deviation, and after the peak, it is accompanied by a large oscillation and drop, corresponding to the current saturation characteristics of the steering actuator. The dotted line of the partially missing control group shows a response pattern between the above two, while the long dashed line of the out-of-range control group shows a peak deviation far exceeding that of the conventional control group and obvious response lag.

[0046] like Fig. 3 As shown, the control algorithm for an autonomous vehicle with emergency obstacle avoidance function includes several key processing stages distributed along the vertical logic main line. The algorithm executes error flux perception, which characterizes transient adjustment pressure by calculating the product of the derivative of the lateral deviation and the rate of change of curvature. It then enters the gain transient attenuation stage, using nonlinear inverse proportional mapping to put the system in a controlled under-gain state to avoid saturation risk. In this process, physical attachment constraints are introduced, and the gain attenuation slope is dynamically limited based on the wheel end slip ratio. Energy smooth recovery is then performed, and the control stiffness is restored stepwise according to the second-order critical damping characteristics. After that, it enters the space safety guidance stage, which guides the margin to the safe area through asymmetric error weight distribution. At the same time, combined with actuator impedance adaptation, the recovery step size is adjusted in real time by injecting micro-disturbance signals. Finally, the load recirculation stage is executed, which maps the control residual into a braking signal to increase the vertical load on the steering wheel.

[0047] Example 4: In an adaptive calibration scenario involving the coordinated alignment of the steering control closed loop and the physical characteristics of the underlying actuators, the system executes a physical linear threshold. The calibration procedure begins when the steering actuator is connected to the controller and the chassis bus is in a statically enabled initial state, where the steering actuator satisfies the requirement that the current control bandwidth is not less than [specified value]. And the position feedback resolution is better than According to the functional specifications, the system injects a ramp-increasing simulated error signal into the steering closed-loop control loop and extracts the drive current of the steering actuator in real time. and the corresponding output torque Monitor drive current Record the critical inflection point that deviates from the linear torque proportional range, and record the error flux corresponding to that inflection point. The value is defined as the hardware saturation threshold, and the hardware saturation threshold is... Determined as a physical linear threshold This calibration procedure correlates the triggering boundaries of the algorithm logic with the physical response capability of the actuator.

[0048] System execution gain adjustment factor The established procedures include collecting lateral deviation data from autonomous vehicles. and the rate of curvature change of the obstacle avoidance target trajectory The first derivative of the lateral deviation was calculated. Calculate the first derivative of the lateral deviation With rate of change of curvature The absolute value of the product determines the error flux. and the error flux With physical linear threshold Perform numerical comparisons; if the error flux... Greater than the physical linear threshold Then the system performs a nonlinear mapping, according to the formula Determine the gain adjustment factor, if the error flux Less than or equal to the physical linear threshold The system will adjust the gain factor. Set as a constant When the vehicle bypasses the obstacle and enters the energy smooth recovery phase, the system uses the identified dynamic impedance... Determine the proportional gain Recovery step size for each control cycle The recovery step size Satisfying the formula: ,in, To satisfy the recovery rate constant of the critical damping constraint, This is the rated mechanical resistance value of the steering actuator. The dynamic impedance is identified in real time through command flow micro-excitation echo analysis. The system uses this formula to calculate the current recovery step size, and adjusts the proportional gain within each control cycle, given the sampling period. Accumulate one recovery step until the proportional gain Restored to rated control stiffness.

[0049] Example 5: In a scenario for parameter calibration of chassis dynamics, the system performs data filling of a nonlinear mapping function and adjustment of the gain factor. The working envelope calibration was performed within a closed test environment, where the ambient temperature was maintained at [temperature value missing]. The actual deflection angle of the steering mechanism is collected by a displacement sensor and compared with the target angle in the command stream. By gradually increasing the frequency of the analog error signal, the error flux at the moment when the output torque of the steering actuator deviates from the linear region is recorded. Numerical values ​​represent the observed error fluxes for each group. With the corresponding proportional gain The attenuation coefficient is fitted using the least squares method, and an error flux is established. Indexed and containing A numerical lookup table for discrete feature points, wherein each entry in the lookup table satisfies that the steering actuator drive current fluctuation rate is less than 1. Physical constraints, thereby establishing error flux on different vehicle hardware platforms. To gain adjustment factor The mapping relationship.

[0050] After the autonomous vehicle completes the replacement of the steering actuator or maintenance of the steering mechanism, the system executes the pre-calibration procedure deployed on-site to establish the proportional gain. The rated mechanical resistance value in the recovery formula The calibration procedure begins when the vehicle is stationary and the steering wheels are at zero offset. The controller injects a frequency of [frequency value missing] into the steering drive motor. And the amplitude is The sinusoidal micro-excitation signal is used to acquire the position signal fed back by the encoder inside the motor, and cross-correlation operation is performed according to the timestamp of the signal stream to extract the phase difference between the position feedback signal and the micro-excitation signal. Based on the response amplitude, the impedance of the current steering mechanism is calculated using the impedance identification formula, and the system executes continuously. The above measurement process is repeated, and the arithmetic mean is taken as the rated mechanical resistance value. The calibration process, which involves writing to non-volatile memory, establishes the correlation between the control compensation step size and the physical state of the hardware. This enables the system to smoothly recover energy after obstacle avoidance based on the real-time identified dynamic impedance. Calculate the proportional gain recovery step size .

[0051] Example 6: In the pre-deployment adaptation procedure for the steering closed-loop system of an autonomous vehicle, the system performs data filling of the nonlinear mapping lookup table and adjustment of the gain factor. The determination of critical constraint feature points, the calibration program starts in the initial state where the steering actuator is unloaded and has rated operating voltage, and the system injects a frequency range from... Increment to And the amplitude is equivalent to The frequency sweep excitation signal for lateral deviation is used to acquire the output torque and phase current of the steering drive motor in real time, and the proportion of harmonic components that cause nonlinear distortion in the torque output is identified. Record the signal characteristics at that time and the corresponding error flux. The value is determined as the maximum index boundary of the lookup table, in the error throughput. from Uniformly select within the interval to the maximum index boundary Each sampling point, and the proportional gain at each sampling point. Conduct critical gain tests to determine that the overshoot of the steering system's step response is no greater than [value missing]. The maximum proportional gain is calculated, and the ratio of this gain to the rated proportional gain is used as the gain adjustment factor. The corresponding entries establish a mapping benchmark between steering control sensitivity and the hardware linear region. After the lookup table is filled, the system performs a recovery rate constant operation. The closed-loop stability calibration procedure uses a dynamometer to simulate the vehicle's self-aligning resistance torque at different driving speeds, and the calibration is performed after the simulated obstacle avoidance maneuver ends and the error flux is measured. Falling back to the physical linear threshold At that moment, the system initiates a gain recovery command stream based on the transient mechanical impedance identification result of the steering actuator, by adjusting the recovery rate constant. Iterate through the values ​​to identify the proportional gain. The signal state in which the rate of change of yaw angular velocity fluctuates the least during the recovery to the rated value is used as the parameter corresponding to the recovery rate constant. And store it in the controller's parameter space, at a vehicle speed of And the lateral deviation is In the obstacle avoidance verification test, the system utilizes the recovery rate constant. Determined recovery step size When the steering mechanism returns to center, the pulse change rate of the steering wheel torque is observed to be within a certain range. The calibration process then achieves dynamic matching between the algorithm adjustment and the chassis mechanical inertia characteristics.

[0052] In spatial constraint adaptation scenarios targeting asymmetric error weight distributions, the system executes the safety-side lateral component. Lateral component of the dangerous side The decomposition and weight allocation, the calibration program starts in the initial state where the controller makes a logical judgment based on the heading angle vector of the vehicle coordinate system and the polar coordinates of the obstacle boundary, and uses the sign function to determine the real-time lateral deviation. Point to the polarity of the safe side or the dangerous side and calculate the lateral component of the safe side. Satisfy the formula The weighting coefficient Set as The system will include the lateral component on the dangerous side. The weighting coefficients are set to constants. The final control deviation signal is synthesized based on the weighted deviation components. The asymmetric distribution of error weights in the spatial dimension induces the displacement margin generated by the steering actuator to be oriented towards the obstacle avoidance safety zone, achieving coordination between flexible control logic and obstacle hard constraints. In calibration scenarios involving the recirculation mapping of control residual energy to chassis load, the system executes the generation procedure of longitudinal load adjustment signal. It is initiated by inputting a steering step signal in a gradient-increasing manner on a smooth road surface. The system extracts the deviation residual amount generated before and after the transient attenuation of the feedback gain in real time, and records the changes in vehicle lateral acceleration fluctuations and front axle vertical load. The system will shorten the convergence time of the center of gravity sideslip angle. The above load increment is determined as the target quantity, and a gain adjustment factor is established. Indexed and containing A lookup table for brake pressure distribution at each feature point, in the case of a vehicle... The moment the vehicle speed enters the obstacle avoidance trajectory is determined based on the currently established gain adjustment factor. The corresponding adjustment command is retrieved from the lookup table, and the electronic braking system is driven to adjust the braking pressure ratio between the front and rear axles from the initial value. Dynamically switch to This induces the vehicle's center of gravity to shift forward and increases the vertical pressure on the steering wheels to improve tire lateral stiffness.

[0053] The embodiments of this application have been described above with reference to the accompanying drawings. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. This application is not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of protection of the technical solutions of this application, and all of these forms are within the protection scope of this application.

Claims

1. A control algorithm for an unmanned vehicle with emergency obstacle avoidance function, characterized in that, Includes the following steps: Step S101: Real-time acquisition of the lateral deviation between the vehicle's current lateral attitude and the obstacle avoidance target trajectory, and determination of the error flux based on the product of the first derivative of the lateral deviation and the rate of change of curvature of the obstacle avoidance target trajectory. Step S102: Input the error flux into a preset nonlinear mapping function to determine the gain adjustment factor; wherein, the nonlinear mapping function limits the gain adjustment factor to have an inverse proportional mapping relationship with the absolute value of the error flux; Step S103: The proportional gain and derivative gain in the steering closed-loop control loop are weighted and weakened in real time using a gain adjustment factor, so that the rate of change of the steering actuator's action command is within the rated physical response envelope of the steering actuator, and the vehicle is in a controlled undergain state. Step S104: Continuously monitor the change in error flux, and determine the preset energy recovery curve when the error flux is less than or equal to the preset physical linearity threshold. Step S105: The proportional gain and differential gain are restored cycle by cycle according to the slope of the energy recovery curve, and the adjusted control quantity is output to drive the steering actuator to perform obstacle avoidance action.

2. The unmanned vehicle control algorithm with emergency obstacle avoidance function according to claim 1, characterized in that, Step S102 specifically includes: calculating the product of the first derivative of the lateral deviation and the rate of change of curvature of the obstacle avoidance target trajectory in real time to determine the adjustment pressure value; mapping the adjustment pressure value to a gain adjustment factor using a preset ramp function; wherein, the gain adjustment factor monotonically decreases as the absolute value of the adjustment pressure value increases; the value range of the gain adjustment factor is limited to between 0.2 and 0.

8.

3. The unmanned vehicle control algorithm with emergency obstacle avoidance function according to claim 1, characterized in that, Also includes: The slip ratio parameters of the vehicle wheel ends are acquired in real time to determine the road adhesion coefficient; the change slope of the gain adjustment factor is dynamically constrained based on the road adhesion coefficient; when the road adhesion coefficient is less than the preset adhesion threshold, the attenuation rate of the proportional gain and the derivative gain is reduced.

4. The unmanned vehicle control algorithm with emergency obstacle avoidance function according to claim 1, characterized in that, Step S103 further includes: determining the safe side orientation of the obstacle avoidance target trajectory relative to the obstacle; decomposing the lateral deviation into the local coordinate system corresponding to the safe side orientation to obtain the safe side lateral component and the dangerous side lateral component; and using a preset weighting coefficient for the safe side lateral component and a normal gain coefficient greater than the weighting coefficient for the dangerous side lateral component to perform the feedback gain reduction action in the direction away from the obstacle.

5. The unmanned vehicle control algorithm with emergency obstacle avoidance function according to claim 1, characterized in that, During the execution of step S103, the method further includes: superimposing a micro-disturbance signal of a preset frequency onto the target control command of the steering actuator; acquiring the position feedback signal of the steering actuator and extracting the response features in the position feedback signal that are at the same frequency as the micro-disturbance signal; and identifying the dynamic impedance of the steering actuator based on the response features. Among them, dynamic impedance Satisfying the formula: ,in, The amplitude of the micro-perturbation signal superimposed on the target control command is denoted as . The response amplitude of the position feedback signal. The phase difference between the position feedback signal and the micro-perturbation signal; utilizing dynamic impedance Adjust the recovery step size of the energy recovery curve.

6. The unmanned vehicle control algorithm with emergency obstacle avoidance function according to claim 1, characterized in that, Also includes: Calculate the residual deviation of the feedback gain before and after adjustment; Map the deviation residual to a longitudinal load adjustment signal; The braking distribution ratio of the front and rear axles of the vehicle is adjusted based on the longitudinal load adjustment signal, thereby increasing the vertical load on the vehicle's steering wheels.

7. The unmanned vehicle control algorithm with emergency obstacle avoidance function according to claim 1, characterized in that, Step S105 specifically includes: monitoring the curvature direction switching point of the obstacle avoidance target trajectory; when reaching the curvature direction switching point, determining the phase compensation pulse based on the current lateral deviation value; and injecting the phase compensation pulse into the steering closed-loop control loop to counteract the vehicle's accumulated lateral kinetic energy.

8. The unmanned vehicle control algorithm with emergency obstacle avoidance function according to claim 1, characterized in that, Step S101 further includes: using a low-pass filter to smooth the lateral deviation and filter out noise components higher than the preset cutoff frequency; the preset cutoff frequency is half of the steering actuator sampling frequency.

9. The unmanned vehicle control algorithm with emergency obstacle avoidance function according to claim 1, characterized in that, After step S105 is executed, the process further includes: extracting the drive current fluctuation of the steering actuator in real time; pausing the recovery process corresponding to the energy recovery curve when the drive current fluctuation is greater than the preset current threshold; maintaining the proportional gain and differential gain at the current level until the drive current fluctuation is less than or equal to the preset current threshold.

10. The unmanned vehicle control algorithm with emergency obstacle avoidance function according to claim 1, characterized in that, The energy recovery curve satisfies the second-order critical damping characteristics; the recovery period of the energy recovery curve is 1.5 to 2.5 times the time required for the steering actuator to complete the rated stroke.

Images