Intelligent automobile active obstacle avoidance control system and method suitable for complex road conditions

Abstract

The invention discloses an intelligent automobile active obstacle avoidance control system and method suitable for complex road conditions. The system comprises a multi-source road condition sensing fusion module, a dynamic risk assessment and decision module, an obstacle avoidance path planning module, a vehicle dynamics control module, a man-machine interaction and emergency takeover module and a system state monitoring and fault diagnosis module. According to the invention, the problem of unreliability of a single sensor in rainy days, congestion and other scenes is solved through multi-source sensor fusion, and the sensing precision is ensured; a multi-factor risk assessment model avoids misjudgment of single distance judgment, and the decision more conforms to actual safety requirements; path planning is combined with vehicle dynamics constraint, closed-loop control ensures accurate path tracking, and the vehicle is prevented from being out of control.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation and vehicle-road cooperative technology, specifically to an intelligent vehicle active obstacle avoidance control system and method applicable to complex road conditions. Background Technology

[0002] With the development of intelligent vehicle technology, active obstacle avoidance has become a core function to ensure driving safety. However, current active obstacle avoidance systems have the following problems:

[0003] A single sensor is prone to failure in severe weather or complex scenarios, and cannot accurately obtain road condition information;

[0004] Judging risk solely by the distance to obstacles, without considering dynamic factors such as obstacle type and relative speed, can easily lead to misjudgment.

[0005] The planned path did not fully consider constraints such as the vehicle's minimum turning radius and lateral acceleration, which could easily lead to loss of vehicle control.

[0006] In cases of system failure or high-risk scenarios, there is no effective takeover mechanism, and drivers are unable to intervene in a timely manner.

[0007] Without establishing redundancy and fault handling mechanisms for sensors and actuators, a single component failure can lead to system failure.

[0008] Therefore, there is an urgent need for an intelligent vehicle active obstacle avoidance control system and method suitable for complex road conditions to solve the problems mentioned above. Summary of the Invention

[0009] The purpose of this invention is to provide an intelligent vehicle active obstacle avoidance control system and method suitable for complex road conditions, so as to solve the problems existing in the prior art.

[0010] To achieve the above objectives, the present invention provides the following technical solution: an intelligent vehicle active obstacle avoidance control system suitable for complex road conditions, including a multi-source road condition perception fusion module, a dynamic risk assessment and decision-making module, an obstacle avoidance path planning module, a vehicle dynamics control module, a human-machine interaction and emergency takeover module, and a system status monitoring and fault diagnosis module.

[0011] The multi-source road condition perception fusion module collects road condition data around the vehicle through multiple types of sensors, and uses a multi-sensor fusion algorithm to achieve redundancy complementarity and accuracy improvement of road condition information, and outputs fusion perception results including obstacle location, type, motion state and road environment characteristics.

[0012] The dynamic risk assessment and decision-making module constructs a dynamic risk assessment model based on the fusion perception results, calculates the comprehensive risk value of the vehicle and surrounding obstacles, and generates corresponding obstacle avoidance decision instructions according to the risk level. The obstacle avoidance decision instructions include deceleration avoidance, steering avoidance, emergency braking and combined obstacle avoidance.

[0013] The obstacle avoidance path planning module generates a smooth and feasible obstacle avoidance path based on the obstacle avoidance decision command and the vehicle's current motion state, which includes speed, steering angle, and acceleration. The vehicle's current motion state includes speed, steering angle, and acceleration. It also combines road constraints and vehicle dynamics constraints, which include lane width and curvature, and vehicle dynamics constraints, including minimum turning radius and maximum lateral acceleration. The module then performs curvature optimization and safety distance verification.

[0014] The vehicle dynamics control module converts the obstacle avoidance path into control signals for the actuators. Through coordinated closed-loop control of the braking system ESP, steering system EPS, and drive system, the vehicle achieves precise tracking of the obstacle avoidance path.

[0015] The human-computer interaction and emergency takeover module provides real-time feedback to the driver on the system's working status, road conditions and risks, and obstacle avoidance actions. It also monitors the driver's operating intentions and issues emergency takeover requests and provides auxiliary information in high-risk or faulty scenarios.

[0016] The system status monitoring and fault diagnosis module monitors the hardware (sensors, processors, actuators) and software status of each module in real time, identifies the fault type and level, and ensures system safety through redundancy replacement, function degradation or emergency handling.

[0017] Preferably, the multi-source road condition perception and fusion module includes a perception data acquisition unit, a data preprocessing unit, a multi-sensor fusion unit, and a perception result output unit;

[0018] The perception data acquisition unit consists of a lidar, a millimeter-wave radar, a vision camera, and an ultrasonic radar, which respectively realize: lidar to acquire the three-dimensional position and contour of obstacles, millimeter-wave radar to acquire the distance / speed / angle of obstacles, vision camera to identify obstacle type / lane lines / traffic signs, and ultrasonic radar to acquire information on obstacles near the vehicle;

[0019] Data preprocessing unit: performs noise reduction, time synchronization, and coordinate transformation on raw data. The LiDAR uses statistical filtering to remove isolated points, and the vision camera uses Gaussian filtering to remove image noise. Time synchronization is based on CAN bus timestamps to align multi-sensor data. Coordinate transformation maps the coordinate systems of each sensor to the vehicle's centroid coordinate system and removes abnormal data.

[0020] Multi-sensor fusion unit: Employs an adaptive weighted fusion algorithm to dynamically adjust sensor weights based on current road conditions, achieving this through three steps: data association, state estimation, and information fusion.

[0021] Data association: Matching the same obstacle detected by different sensors using the nearest neighbor algorithm;

[0022] State estimation: Kalman filtering is used to dynamically predict and update the motion state of obstacles;

[0023] Information fusion: Weighting multi-source data with the same state variable according to weights and outputting the fusion result;

[0024] Perception Result Output Unit: Transmits the fused obstacle information and road environment information to subsequent modules in a preset format. The obstacle information includes the movement trend of position, speed, and category. The road environment information includes lane line position and road boundary. The data is stored for backtracking.

[0025] Preferably, the dynamic risk assessment and decision-making module includes an obstacle classification unit, a risk calculation unit, a decision generation unit, and a decision verification unit;

[0026] Obstacle classification unit: Based on a deep learning recognition algorithm using a visual camera and combined with the reflection characteristics of millimeter-wave radar, obstacles are classified into four categories: pedestrians, non-motorized vehicles, motorized vehicles, and static obstacles, providing a basis for type weighting for risk calculation;

[0027] Risk Calculation Unit: Constructs a comprehensive risk assessment model, using the risk value calculation formula:

[0028]

[0029] in, This is the combined risk value of the vehicle and obstacles, ranging from 0 to 1, with a higher value indicating a higher risk. The shortest relative distance between the vehicle's center of mass and the obstacle's center of mass. >0, the smaller the distance, the greater the risk contribution; The relative velocity between the vehicle and the obstacle, when the obstacle moves toward the vehicle. A positive value indicates that the greater the relative speed, the greater the risk contribution. This is the distance weighting coefficient, ranging from 0.4 to 0.6, reflecting the degree of influence of distance on risk; The relative speed weighting coefficient, with a value of 0.3-0.5, reflects the degree of influence of relative speed on risk. This is the obstacle type weighting coefficient, ranging from 0.1 to 0.2, reflecting the degree of impact of the type on the risk; For obstacle type coefficients, pedestrians =3. Non-motorized vehicles =2. Motor vehicles =1. Static obstacles =0.5, distinguishing the danger levels of different obstacles;

[0030] Decision generation unit: Classifies risk levels based on risk value R:

[0031] Low risk <0.3: Maintain current driving direction;

[0032] Medium risk 0.3≤ <0.7: Deceleration / Steering to avoid an obstacle;

[0033] High risk ≥0.7: Emergency braking / combined obstacle avoidance; generates preliminary decisions based on road conditions and vehicle status;

[0034] Decision verification unit: Verifies the feasibility of the initial decision (e.g., when turning to avoid obstacles, it is necessary to confirm that there are no obstacles in the adjacent lanes; when braking urgently, it is necessary to confirm that the braking distance is less than or equal to the current relative distance). If it is not feasible, the decision is adjusted (e.g., if turning is not possible, it is switched to emergency braking), and finally outputs the obstacle avoidance decision command.

[0035] Preferably, the obstacle avoidance path planning module includes a constraint determination unit, a path generation unit, a path optimization unit, and a path verification unit;

[0036] Constraint determination unit: Define three types of constraints:

[0037] Vehicle dynamics constraints: minimum turning radius Maximum lateral acceleration ;

[0038] Road constraints: Lane boundaries must not cross the line; road curvature must be matched.

[0039] Safety constraint: The minimum safe distance between the path and obstacles is greater than or equal to the preset value;

[0040] Path generation unit: based on obstacle avoidance decision-making algorithm:

[0041] Deceleration and avoidance → Straight deceleration path: Maintain the current lane and plan a straight line from the current position to the deceleration target position;

[0042] Steering and obstacle avoidance → Third-order Bézier curve path, using the path generation formula:

[0043]

[0044] in, for Point coordinates on the time path That is, the target location of the vehicle; For path parameters, =0 corresponds to the starting point of the path. Current centroid coordinates of the vehicle , =1 corresponds to the end point of the path ; , As an intermediate control point: Based on the vehicle's current direction of travel, it is located within the current lane and... The distance is a safe distance. Based on the direction of the route's endpoint and the road boundary, it is located within the target lane, and... The distance is a safe distance;

[0045] Path optimization unit:

[0046] Curvature optimization: Calculate the curvature at each point along the path ,make sure ≤1 / (Meets minimum turning radius), adjustments made when exceeding limits. , Location;

[0047] Smoothing optimization: Sliding window filtering is used to smooth path points and reduce steering shock;

[0048] Path verification unit: Calculates the minimum distance between the optimized path and obstacles, verifies whether the safety constraints are met, and readjusts the control points until the path is feasible.

[0049] Preferably, the vehicle dynamics control module includes a control target conversion unit, a steering control unit, a braking / drive control unit, and a closed-loop feedback unit;

[0050] Control target conversion unit: converts the location information of the obstacle avoidance path. Transformed into objectives of implementing agencies:

[0051] Steering target: Front wheel target steering angle at various times ;

[0052] Speed ​​target: Target speed at each moment ;

[0053] Steering control unit: Employs a PID control algorithm to calculate the output torque of the steering motor.

[0054]

[0055] in, The actual steering angle fed back by the steering angle sensor is used to drive the front wheels to track the steering wheel via the EPS system. ;

[0056] Braking / Drive Control Unit: Employs a sliding mode control algorithm, based on the target speed. Calculate the target deceleration , The distance from the current location to the target location is used to drive the ESP system or the motor / engine to adjust the actual speed. track ;

[0057] Closed-loop feedback unit: Real-time acquisition of wheel speed sensor data Steering angle sensor Accelerometer Use the data to adjust the control parameters and ensure control accuracy.

[0058] Preferably, the human-computer interaction and emergency takeover module includes a status feedback unit, an operation intention monitoring unit, and an emergency takeover unit;

[0059] Status feedback unit: Displays system status, risk information, and obstacle avoidance parameters on the instrument panel and central control screen, and provides voice prompts;

[0060] The driver intention monitoring unit collects driver input via steering wheel angle sensor and accelerator / brake pedal sensor. If a large steering input or deep pedal depress is detected, it determines that the driver intends to take over and gradually reduces the system's control authority.

[0061] Emergency takeover unit: When a high risk is detected ( When the error is ≥0.9 and the path cannot be avoided, or when there is a system malfunction, an audible and visual alarm will be issued to prompt the driver to take over; takeover assistance information will be provided; if the driver does not take over within the preset time, the backup brake will be activated.

[0062] Preferably, the system status monitoring and fault diagnosis module includes a status acquisition unit, a fault identification unit, and a fault processing unit;

[0063] Status acquisition unit: collects data from each module in real time.

[0064] Sensor status, including data packet loss rate and transmission latency;

[0065] Processor status, including CPU utilization and memory usage;

[0066] Actuator status, including braking pressure and steering motor current;

[0067] Communication status, including CAN bus bit error rate;

[0068] Fault identification unit: It adopts an algorithm that combines fault tree analysis (FTA) with neural networks to identify fault types (sensor faults, communication faults, actuator faults) and levels: minor faults, moderate faults, and severe faults;

[0069] Fault handling unit:

[0070] Minor malfunction: Replace with redundant sensors and obstacle avoidance will continue normally;

[0071] Moderate fault: Adjust algorithm parameters, limit the vehicle's maximum speed, and remind the driver to be cautious;

[0072] Major malfunction: Immediately trigger emergency takeover, activate backup systems, and ensure vehicle safety.

[0073] A control method for an intelligent vehicle active obstacle avoidance control system suitable for complex road conditions includes the following steps:

[0074] Step 1: System Initialization and Sensing Calibration

[0075] Start the obstacle avoidance system and complete sensor calibration; monitor the initial status of hardware / software, enter standby if there are no faults, and prompt for maintenance if there are faults;

[0076] Step 2: Multi-source sensing data acquisition and fusion

[0077] Data is collected from lidar, millimeter-wave radar, vision camera, and ultrasonic radar. After denoising, time synchronization, and coordinate transformation of the data, the sensor weights are dynamically adjusted according to road conditions. Data fusion is completed through data association and Kalman filtering, and road condition information is output to the risk assessment module.

[0078] Step 3: Dynamic Risk Assessment and Obstacle Avoidance Decisions

[0079] Identify obstacle types and determine type coefficients; substitute them into the risk formula to calculate the overall risk value. Classify risk levels; generate preliminary obstacle avoidance decisions; and output instructions after verifying feasibility.

[0080] Step 4: Obstacle Avoidance Path Planning and Optimization

[0081] Define vehicle dynamics, road conditions, and safety constraints; generate a path based on the decision; optimize path curvature and smoothness; verify safe distances and then transmit the results to the control module.

[0082] Step 5: Vehicle dynamics control execution

[0083] The path is converted into target values ​​for steering angle and speed; the steering system is controlled by a PID algorithm and the braking / drive system is controlled by a sliding mode algorithm to track the target values; the actual speed and steering angle are collected, and the control quantity is adjusted in a closed loop to ensure accuracy.

[0084] Step 6: Human-Computer Interaction Feedback and Emergency Takeover

[0085] The system provides status and risk information via display screen / voice feedback; it monitors driver actions and gradually transfers control upon detecting active actions; high-risk situations require close monitoring. If the value is ≥0.9 or there is a fault, an audible and visual alarm will prompt the operator to take over. If the operator does not take over within 3 seconds, the backup brake will be activated.

[0086] Step 7: System Status Monitoring and Fault Handling

[0087] It collects real-time data on the status of sensors, processors, actuators, and communication; identifies fault types and levels and handles them accordingly; and records obstacle avoidance data for subsequent optimization.

[0088] Step 8: Obstacle avoidance completed and system reset

[0089] Risk reduced to When the value is less than 0.3, adjust the vehicle to the road speed limit and straighten the steering wheel; determine that the risk has been eliminated and notify the driver that obstacle avoidance has been completed; reset all modules to standby, waiting for the next trigger.

[0090] Compared with the prior art, the beneficial effects of the present invention are:

[0091] This invention addresses the unreliability of single sensors in scenarios such as rain and traffic congestion through multi-source sensor fusion, ensuring perception accuracy. A multi-factor risk assessment model avoids misjudgments based on single distance assessments, making decisions more aligned with actual safety requirements. Path planning combined with vehicle dynamics constraints and closed-loop control ensure accurate path tracking and prevent vehicle loss of control. Real-time feedback and emergency takeover mechanisms balance system autonomy with driver intervention needs, reducing risk. Hierarchical fault handling and redundancy design ensure that minor faults do not affect obstacle avoidance, while severe faults guarantee vehicle safety. Attached Figure Description

[0092] Figure 1 This is a system module diagram of the present invention;

[0093] Figure 2 This is a schematic diagram of the multi-source road condition perception fusion module of the present invention;

[0094] Figure 3 This is a schematic diagram of the dynamic risk assessment and decision-making module of the present invention;

[0095] Figure 4 This is a flowchart of the method of the present invention. Detailed Implementation

[0096] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0097] Please see Figure 1-4This invention provides an intelligent vehicle active obstacle avoidance control system suitable for complex road conditions, including a multi-source road condition perception fusion module, a dynamic risk assessment and decision-making module, an obstacle avoidance path planning module, a vehicle dynamics control module, a human-machine interaction and emergency takeover module, and a system status monitoring and fault diagnosis module.

[0098] The multi-source road condition perception fusion module collects road condition data around the vehicle through multiple types of sensors, and uses a multi-sensor fusion algorithm to achieve redundancy complementarity and accuracy improvement of road condition information, and outputs fusion perception results including obstacle location, type, motion state and road environment characteristics.

[0099] The dynamic risk assessment and decision-making module constructs a dynamic risk assessment model based on the fusion perception results, calculates the comprehensive risk value of the vehicle and surrounding obstacles, and generates corresponding obstacle avoidance decision instructions according to the risk level. The obstacle avoidance decision instructions include deceleration avoidance, steering avoidance, emergency braking and combined obstacle avoidance.

[0100] The obstacle avoidance path planning module generates a smooth and feasible obstacle avoidance path based on the obstacle avoidance decision command and the vehicle's current motion state, which includes speed, steering angle, and acceleration. The vehicle's current motion state includes speed, steering angle, and acceleration. It also combines road constraints and vehicle dynamics constraints, which include lane width and curvature, and vehicle dynamics constraints, including minimum turning radius and maximum lateral acceleration. The module then performs curvature optimization and safety distance verification.

[0101] The vehicle dynamics control module converts the obstacle avoidance path into control signals for the actuators. Through coordinated closed-loop control of the braking system (ESP), steering system (EPS), and drive system, it enables the vehicle to accurately track the obstacle avoidance path.

[0102] The human-computer interaction and emergency takeover module provides real-time feedback to the driver on the system's operating status, road conditions and risks, and obstacle avoidance actions. It also monitors the driver's operating intentions and issues emergency takeover requests and provides auxiliary information in high-risk or faulty scenarios.

[0103] The system status monitoring and fault diagnosis module monitors the hardware (sensors, processors, actuators) and software status of each module in real time, identifies the type and level of faults, and ensures system safety through redundancy replacement, function degradation or emergency handling.

[0104] The multi-source road condition perception and fusion module includes a perception data acquisition unit, a data preprocessing unit, a multi-sensor fusion unit, and a perception result output unit.

[0105] The perception data acquisition unit consists of a lidar, a millimeter-wave radar, a vision camera, and an ultrasonic radar, which respectively realize: lidar to acquire the three-dimensional position and contour of obstacles, millimeter-wave radar to acquire the distance / speed / angle of obstacles, vision camera to identify obstacle type / lane lines / traffic signs, and ultrasonic radar to acquire information on obstacles near the vehicle;

[0106] Data preprocessing unit: performs noise reduction, time synchronization, and coordinate transformation on raw data. The LiDAR uses statistical filtering to remove isolated points, and the vision camera uses Gaussian filtering to remove image noise. Time synchronization is based on CAN bus timestamps to align multi-sensor data. Coordinate transformation maps the coordinate systems of each sensor to the vehicle's centroid coordinate system and removes abnormal data.

[0107] Multi-sensor fusion unit: Employs an adaptive weighted fusion algorithm to dynamically adjust sensor weights based on current road conditions, achieving this through three steps: data association, state estimation, and information fusion.

[0108] Data association: Matching the same obstacle detected by different sensors using the nearest neighbor algorithm;

[0109] State estimation: Kalman filtering is used to dynamically predict and update the motion state of obstacles;

[0110] Information fusion: Weighting multi-source data with the same state variable according to weights and outputting the fusion result;

[0111] Perception Result Output Unit: Transmits the fused obstacle information and road environment information to subsequent modules in a preset format. The obstacle information includes the movement trend of position, speed, and category. The road environment information includes lane line position and road boundary. The data is stored for backtracking.

[0112] The dynamic risk assessment and decision-making module includes an obstacle classification unit, a risk calculation unit, a decision generation unit, and a decision verification unit.

[0113] Obstacle classification unit: Based on a deep learning recognition algorithm using a visual camera and combined with the reflection characteristics of millimeter-wave radar, obstacles are classified into four categories: pedestrians, non-motorized vehicles, motorized vehicles, and static obstacles, providing a basis for type weighting for risk calculation;

[0114] Risk Calculation Unit: Constructs a comprehensive risk assessment model, using the risk value calculation formula:

[0115]

[0116] in, This is the combined risk value of the vehicle and obstacles, ranging from 0 to 1, with a higher value indicating a higher risk. The shortest relative distance between the vehicle's center of mass and the obstacle's center of mass. >0, the smaller the distance, the greater the risk contribution; The relative velocity between the vehicle and the obstacle, when the obstacle moves toward the vehicle. A positive value indicates that the greater the relative speed, the greater the risk contribution. This is the distance weighting coefficient, ranging from 0.4 to 0.6, reflecting the degree of influence of distance on risk; The relative speed weighting coefficient, with a value of 0.3-0.5, reflects the degree of influence of relative speed on risk. This is the obstacle type weighting coefficient, ranging from 0.1 to 0.2, reflecting the degree of impact of the type on the risk; For obstacle type coefficients, pedestrians =3. Non-motorized vehicles =2. Motor vehicles =1. Static obstacles =0.5, distinguishing the danger levels of different obstacles;

[0117] Decision generation unit: Classifies risk levels based on risk value R:

[0118] Low risk <0.3: Maintain current driving direction;

[0119] Medium risk 0.3≤ <0.7: Deceleration / Steering to avoid an obstacle;

[0120] High risk ≥0.7: Emergency braking / combined obstacle avoidance; generates preliminary decisions based on road conditions and vehicle status;

[0121] Decision verification unit: Verifies the feasibility of the initial decision (e.g., when turning to avoid obstacles, it is necessary to confirm that there are no obstacles in the adjacent lanes; when braking urgently, it is necessary to confirm that the braking distance is less than or equal to the current relative distance). If it is not feasible, the decision is adjusted (e.g., if turning is not possible, it is switched to emergency braking), and finally outputs the obstacle avoidance decision command.

[0122] The obstacle avoidance path planning module includes a constraint determination unit, a path generation unit, a path optimization unit, and a path verification unit.

[0123] Constraint determination unit: Define three types of constraints:

[0124] Vehicle dynamics constraints: minimum turning radius Maximum lateral acceleration ;

[0125] Road constraints: Lane boundaries must not cross the line; road curvature must be matched.

[0126] Safety constraint: The minimum safe distance between the path and obstacles is greater than or equal to the preset value;

[0127] Path generation unit: based on obstacle avoidance decision-making algorithm:

[0128] Deceleration and avoidance → Straight deceleration path: Maintain the current lane and plan a straight line from the current position to the deceleration target position;

[0129] Steering and obstacle avoidance → Third-order Bézier curve path, using the path generation formula:

[0130]

[0131] in, for Point coordinates on the time path That is, the target location of the vehicle; For path parameters, =0 corresponds to the starting point of the path. Current centroid coordinates of the vehicle , =1 corresponds to the end point of the path ; , As an intermediate control point: Based on the vehicle's current direction of travel, it is located within the current lane and... The distance is a safe distance. Based on the direction of the route's endpoint and the road boundary, it is located within the target lane, and... The distance is a safe distance;

[0132] Path optimization unit:

[0133] Curvature optimization: Calculate the curvature at each point along the path ,make sure ≤1 / (Meets minimum turning radius), adjustments made when exceeding limits. , Location;

[0134] Smoothing optimization: Sliding window filtering is used to smooth path points and reduce steering shock;

[0135] Path verification unit: Calculates the minimum distance between the optimized path and obstacles, verifies whether the safety constraints are met, and readjusts the control points until the path is feasible.

[0136] The vehicle dynamics control module includes a control target conversion unit, a steering control unit, a braking / drive control unit, and a closed-loop feedback unit;

[0137] Control target conversion unit: converts the location information of the obstacle avoidance path. Transformed into objectives of implementing agencies:

[0138] Steering target: Front wheel target steering angle at various times ;

[0139] Speed ​​target: Target speed at each moment ;

[0140] Steering control unit: Employs a PID control algorithm to calculate the output torque of the steering motor.

[0141]

[0142] in, The actual steering angle fed back by the steering angle sensor is used to drive the front wheels to track the steering wheel via the EPS system. ;

[0143] Braking / Drive Control Unit: Employs a sliding mode control algorithm, based on the target speed. Calculate the target deceleration , The distance from the current location to the target location is used to drive the ESP system or the motor / engine to adjust the actual speed. track ;

[0144] Closed-loop feedback unit: Real-time acquisition of wheel speed sensor data Steering angle sensor Accelerometer Use the data to adjust the control parameters and ensure control accuracy.

[0145] The human-computer interaction and emergency takeover module includes a status feedback unit, an operation intent monitoring unit, and an emergency takeover unit;

[0146] Status feedback unit: Displays system status, risk information, and obstacle avoidance parameters on the instrument panel and central control screen, and provides voice prompts;

[0147] The driver intention monitoring unit collects driver input via steering wheel angle sensor and accelerator / brake pedal sensor. If a large steering input or deep pedal depress is detected, it determines that the driver intends to take over and gradually reduces the system's control authority.

[0148] Emergency takeover unit: When a high risk is detected ( When the error is ≥0.9 and the path cannot be avoided, or when there is a system malfunction, an audible and visual alarm will be issued to prompt the driver to take over; takeover assistance information will be provided; if the driver does not take over within the preset time, the backup brake will be activated.

[0149] The system status monitoring and fault diagnosis module includes a status acquisition unit, a fault identification unit, and a fault processing unit;

[0150] Status acquisition unit: collects data from each module in real time.

[0151] Sensor status, including data packet loss rate and transmission latency;

[0152] Processor status, including CPU utilization and memory usage;

[0153] Actuator status, including braking pressure and steering motor current;

[0154] Communication status, including CAN bus bit error rate;

[0155] Fault identification unit: It adopts an algorithm that combines fault tree analysis (FTA) with neural networks to identify fault types (sensor faults, communication faults, actuator faults) and levels: minor faults, moderate faults, and severe faults;

[0156] Fault handling unit:

[0157] Minor malfunction: Replace with redundant sensors and obstacle avoidance will continue normally;

[0158] Moderate fault: Adjust algorithm parameters, limit the vehicle's maximum speed, and remind the driver to be cautious;

[0159] Major malfunction: Immediately trigger emergency takeover, activate backup systems, and ensure vehicle safety.

[0160] A control method for an intelligent vehicle active obstacle avoidance control system suitable for complex road conditions includes the following steps:

[0161] Step 1: System Initialization and Sensing Calibration

[0162] The active obstacle avoidance system is activated, and each module performs initialization:

[0163] Multi-source road condition perception fusion module: calibrates sensors (Coordinate system 1: maps each sensor to the vehicle's center of gravity coordinate system; Time synchronization: ensures data acquisition time deviation ≤ threshold; Parameter correction: eliminates inherent sensor errors).

[0164] System status monitoring and fault diagnosis module: detects the initial status of hardware (sensors, processor) and software (algorithm, storage). If there is no fault, it enters standby mode; if there is a fault, it prompts for maintenance.

[0165] Step 2: Multi-source sensing data acquisition and fusion

[0166] Data Acquisition: The acquisition unit of the multi-source sensing module acquires road condition data in real time: LiDAR acquires the three-dimensional position of obstacles, millimeter-wave radar acquires distance / velocity, visual camera acquires road images, and ultrasonic radar acquires nearby obstacles;

[0167] Data preprocessing: The preprocessing unit denoises the raw data (statistical filtering for LiDAR, Gaussian filtering for vision camera), synchronizes the time (timestamp alignment for CAN bus), transforms the coordinates (unifying to the vehicle's centroid coordinate system), and removes outliers;

[0168] Multi-sensor fusion: The fusion unit adjusts the sensor weights according to the current road conditions, matches the same obstacle through data association, estimates the obstacle's motion state using Kalman filtering, and outputs the fusion result (obstacle position / speed / category, road environment), which is then transmitted to the risk assessment module.

[0169] Step 3: Dynamic Risk Assessment and Obstacle Avoidance Decisions

[0170] Obstacle classification: The classification unit of the risk assessment module identifies obstacle types and determines the type coefficient c;

[0171] Risk calculation: Substitute the risk calculation unit into the formula Calculate the risk value of each obstacle. ;

[0172] Risk classification: based on Classify levels and generate preliminary decisions;

[0173] Decision verification: The verification unit confirms the feasibility of the decision, adjusts it, and outputs obstacle avoidance decision instructions (including target speed and turning angle), which are then transmitted to the path planning and human-computer interaction module.

[0174] Step 4: Obstacle Avoidance Path Planning and Optimization

[0175] Constraints are defined: The constraint unit of the path planning module specifies three types of constraints: vehicle dynamics (minimum turning radius), road (lane boundaries), and safety (safe distance from obstacles).

[0176] Path generation: Based on the decision-making algorithm: Bessel formula is used for steering and obstacle avoidance. Generate path, determine , , / ;

[0177] Path optimization: Optimize the calculation of path curvature in the optimization unit and adjust... / Ensure curvature ≤ 1 / Sliding window filtering smooths the path;

[0178] Path verification: Verify that the minimum distance between the path and obstacles is greater than or equal to the safety value. If it is feasible, the path is transmitted to the dynamics control module; otherwise, the path is replanned.

[0179] Step 5: Vehicle dynamics control execution

[0180] Target transformation: The dynamics control module transforms the path into an execution target, including the steering angle. ,speed ;

[0181] Steering control: The steering unit uses a PID algorithm to calculate the steering torque, driving the EPS system to adjust the front wheels and track the steering wheel. ;

[0182] Braking / Drive Control: The braking / drive unit uses a sliding mode algorithm to calculate braking pressure / drive torque, drives the ESP / motor, and tracks... ;

[0183] Closed-loop feedback: The feedback unit collects the actual speed. Steering angle Adjust the control parameters to ensure path tracking accuracy.

[0184] Step 6: Human-Computer Interaction Feedback and Emergency Takeover

[0185] Status feedback: The human-machine interaction module provides the driver with feedback on system status and risk information via display screen / voice.

[0186] Intent monitoring: Monitor driver actions; if active actions are detected, gradually transfer control.

[0187] Emergency takeover: If risks If the error is ≥0.9 or there is a system malfunction, an audible and visual alarm will be issued to prompt the driver to take over. If the driver does not take over within 3 seconds, the backup brake will be activated until the vehicle stops.

[0188] Step 7: System Status Monitoring and Fault Handling

[0189] Status Acquisition: The fault diagnosis module collects the status of each module in real time;

[0190] Fault identification: Identify the type and severity of faults;

[0191] Fault handling: Minor faults → redundancy replacement; Moderate faults → parameter adjustment + speed limitation; Severe faults → emergency takeover + backup system;

[0192] Data logging: Records the obstacle avoidance data for subsequent algorithm optimization.

[0193] Step 8: Obstacle avoidance completed and system reset

[0194] When the vehicle avoids the obstacle, the risk <0.3, return to normal driving status:

[0195] The dynamics control module adjusts the vehicle to the road speed limit and the steering returns to center.

[0196] The risk assessment module determines that the risk has been eliminated and generates an obstacle avoidance completion command.

[0197] The human-machine interface module indicated to the driver that obstacle avoidance was complete, and the system returned to normal.

[0198] Each module is reset to standby mode, awaiting the next obstacle avoidance trigger.

[0199] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An intelligent automobile active obstacle avoidance control system suitable for complex road conditions, characterized in that: The system comprises a multi-source road condition perception fusion module, a dynamic risk assessment and decision module, an obstacle avoidance path planning module, a vehicle dynamics control module, a human-machine interaction and emergency takeover module, and a system state monitoring and fault diagnosis module. The multi-source road condition perception fusion module collects vehicle surrounding road condition data through multiple types of sensors, uses a multi-sensor fusion algorithm to realize redundancy complementation and precision improvement of road condition information, and outputs fusion perception results containing obstacle position, category, motion state and road environment features. The dynamic risk assessment and decision module constructs a dynamic risk assessment model based on the fusion perception results, calculates the comprehensive risk value of the vehicle and the surrounding obstacles, and generates corresponding obstacle avoidance decision instructions according to the risk level, including deceleration avoidance, turning avoidance, emergency braking and combined obstacle avoidance. The obstacle avoidance path planning module generates a smooth and feasible obstacle avoidance path according to the obstacle avoidance decision instructions and the current motion state of the vehicle, including speed, steering angle and acceleration, and combines road constraints including lane width and curvature, and vehicle dynamics constraints including minimum turning radius and maximum lateral acceleration, and performs curvature optimization and safety distance verification. The vehicle dynamics control module converts the obstacle avoidance path into actuator control signals, and performs collaborative closed-loop control on the brake system ESP, steering system EPS and driving system. The human-machine interaction and emergency takeover module feeds back the system working state, road condition risk and obstacle avoidance action to the driver in real time, monitors the driver's operation intention, issues an emergency takeover request and provides auxiliary information in high-risk or fault scenarios. The system state monitoring and fault diagnosis module monitors the hardware and software states of each module in real time, identifies the fault type and level, and ensures system safety through redundancy replacement, function degradation or emergency treatment.

2. The active obstacle avoidance control system for intelligent vehicle suitable for complex road conditions as claimed in claim 1 wherein: The multi-source road condition perception fusion module comprises a perception data acquisition unit, a data preprocessing unit, a multi-sensor fusion unit and a perception result output unit. The perception data acquisition unit comprises a laser radar, a millimeter wave radar, a vision camera and an ultrasonic radar, which respectively realize laser radar acquisition of obstacle three-dimensional position and contour, millimeter wave radar acquisition of obstacle distance / speed / angle, vision camera identification of obstacle category / lane line / traffic sign, and ultrasonic radar acquisition of vehicle close-range obstacle information. The data preprocessing unit performs denoising, time synchronization and coordinate conversion on the original data, uses statistical filtering to remove isolated points for the laser radar, uses Gaussian filtering to remove image noise for the vision camera, aligns multi-sensor data based on CAN bus time stamp for time synchronization, and maps each sensor coordinate system to the vehicle mass center coordinate system for coordinate conversion and removal of abnormal data. The multi-sensor fusion unit uses an adaptive weighted fusion algorithm to dynamically adjust the sensor weight according to the current road condition, and realizes data association-state estimation-information fusion in three steps: Data association: uses the nearest neighbor algorithm to match the same obstacle detected by different sensors; State estimation: uses Kalman filtering to dynamically predict and update the motion state of the obstacle; Information fusion: multiple source data of the same state quantity are weighted according to weights, and a fusion result is output; A perception result output unit: the fused obstacle information and road environment information are transmitted to subsequent modules in a preset format, the obstacle information includes position, speed, and category of motion trend, and the road environment information includes lane line position and road boundary, and the data is stored for backtracking.

3. The active obstacle avoidance control system for intelligent vehicle suitable for complex road conditions as claimed in claim 1 wherein: The dynamic risk assessment and decision module includes an obstacle classification unit, a risk calculation unit, a decision generation unit, and a decision verification unit; The obstacle classification unit: based on a deep learning recognition algorithm of a visual camera, combined with the reflection characteristics of a millimeter wave radar, the obstacles are classified into four categories: pedestrians, non-motor vehicles, motor vehicles, and static obstacles, providing type weight basis for risk calculation; The risk calculation unit: a comprehensive risk assessment model is constructed, and a risk value calculation formula is adopted: ； wherein, is the comprehensive risk value of the vehicle and the obstacle, with a value range of 0-1, and the greater the value, the higher the risk; is the shortest relative distance between the vehicle centroid and the obstacle centroid, > 0, the smaller the distance, the greater the risk contribution; is the relative speed between the vehicle and the obstacle, and when the obstacle moves towards the vehicle is positive, the greater the relative speed, the greater the risk contribution; is the distance weight coefficient, with a value of 0.4-0.6, reflecting the degree of influence of distance on risk; is the relative speed weight coefficient, with a value of 0.3-0.5, reflecting the degree of influence of relative speed on risk; is the obstacle type weight coefficient, with a value of 0.1-0.2, reflecting the degree of influence of type on risk; is the obstacle type coefficient, pedestrians = 3, non-motor vehicles = 2, motor vehicles = 1, static obstacles = 0.5, to distinguish the risk levels of different obstacles; The decision generation unit: according to the risk value R, the risk level is divided: Low risk <0.3: keep current travel; Medium risk 0.3 < x < 0.5 <0.7: decelerate / swerve to avoid; High risk ≥ 0.7: emergency braking / combined obstacle avoidance; preliminary decision generated combining road conditions and vehicle state; The decision verification unit: the feasibility of the preliminary decision is verified, and if it is not feasible, the decision is adjusted, and finally the obstacle avoidance decision instruction is output.

4. The active obstacle avoidance control system for intelligent vehicle suitable for complex road conditions as claimed in claim 1 wherein: The obstacle avoidance path planning module includes a constraint condition determination unit, a path generation unit, a path optimization unit, and a path verification unit; The constraint condition determination unit: three types of constraints are defined: Vehicle dynamics constraints: minimum turning radius , maximum lateral acceleration ; Road constraint: lane boundary does not cross the line, road curvature matching; Safety constraint: the minimum safety distance between the path and the obstacle is greater than a preset value; The path generation unit: according to the obstacle avoidance decision selection algorithm: Speed reduction avoidance→straight line speed reduction path: maintain the current lane and plan a straight line from the current position to the speed reduction target position; Steering avoidance→three-order Bezier curve path, using the path generation formula: ； wherein, is the point coordinate on the path at the moment , i.e. the vehicle target position; is the path parameter, =0 corresponds to the path start point is the vehicle current centroid coordinate , =1 corresponds to the path end point ; , is the intermediate control point: according to the vehicle current driving direction setting, located in the current lane, and the distance is a safe distance, according to the path end point direction and the road boundary setting, located in the target lane, and the distance is a safe distance; The path optimization unit: Curvature optimization: calculate curvature of each point of path , ensure ≤1 / , adjust when over limit , position; Smooth optimization: use a sliding window filter to smooth the path points to reduce the steering impact; The path verification unit: calculate the minimum distance between the optimized path and the obstacle to verify whether the safety constraint is met, if not, adjust the control points until the path is feasible.

5. The active obstacle avoidance control system for intelligent vehicle suitable for complex road conditions as claimed in claim 1 wherein: The vehicle dynamics control module includes a control target conversion unit, a steering control unit, a brake / drive control unit, and a closed-loop feedback unit; Control target conversion unit: converts the position information of the obstacle avoidance path into an actuator target Steering target: front wheel target steering angle at each time ; Speed target: target speed at each time ; The steering control unit: adopts a PID control algorithm to calculate the output torque of the steering motor: ； wherein, is the actual steering angle fed back by the steering angle sensor, the front wheel is tracked by the EPS system driven ; Brake / drive control unit: uses a sliding mode control algorithm to determine the target speed Calculates target deceleration , Distance to target position, drives the ESP system or the motor / engine to make the actual speed Follows the target speed ; Closed loop feedback unit: real-time acquisition of wheel speed sensor , steering angle sensor , acceleration sensor Data, adjustment control, ensure control accuracy.

6. The active obstacle avoidance control system for intelligent vehicle suitable for complex road conditions as claimed in claim 1 wherein: The human-machine interaction and emergency takeover module includes a state feedback unit, an operation intention monitoring unit, and an emergency takeover unit; The state feedback unit: displays the system state, risk information, and obstacle avoidance parameters through the instrument panel and central control screen, and also provides voice prompts; The operation intention monitoring unit: collects the driver's operation through the steering wheel angle sensor and the accelerator / brake pedal sensor, and if a large steering or deep pedal is detected, it is determined that the driver intends to take over, and the system control authority is gradually reduced; The emergency takeover unit: when high risk or system failure is detected, sound and light alarms are issued to prompt the driver to take over; Provide takeover assistance information; If the takeover is not completed within a preset time, the backup brake is started.

7. The active obstacle avoidance control system for intelligent vehicle suitable for complex road conditions as claimed in claim 1 wherein: The system state monitoring and fault diagnosis module includes a state acquisition unit, a fault identification unit, and a fault handling unit; The state acquisition unit: real-time acquisition of module data: Sensor state, including data packet loss rate and transmission delay; Processor state, including CPU occupancy and memory usage; Actuator state, including brake pressure and steering motor current; Communication status, including CAN bus error rate; Fault identification unit: using fault tree analysis (FTA) and neural network combined algorithm to identify fault type and level: mild fault, moderate fault and severe fault; Fault handling unit: Mild fault: replaced by redundant sensor, normal execution of obstacle avoidance; Moderate fault: adjust algorithm parameters, limit vehicle speed, prompt driver to be cautious; Severe fault: immediately trigger emergency takeover, start backup system to ensure vehicle safety. 8.The control method of the active obstacle avoidance control system for intelligent vehicle suitable for complex road conditions according to any one of claims 1-7, characterized in that: Including the following steps: Step 1: system initialization and perception calibration Start obstacle avoidance system, complete sensor calibration; Monitor the initial state of hardware / software, no fault into standby, fault prompt maintenance; Step 2: multi-source perception data acquisition and fusion Collect laser radar, millimeter wave radar, vision camera, ultrasonic radar data; After denoising, time synchronization, coordinate conversion, adjust sensor weight according to road conditions, complete fusion through data association and Kalman filtering, output road condition information to risk assessment module; Step 3: dynamic risk assessment and obstacle avoidance decision Identify the type of obstacle, determine the type coefficient; Substitute the risk formula to calculate the comprehensive risk value , divide the risk level; Generate preliminary obstacle avoidance decision, verify feasibility and output instruction; Step 4: obstacle avoidance path planning and optimization Determine vehicle dynamics, road, safety constraints; Generate path according to decision; Optimize path curvature and smoothness, verify safety distance and transmit to control module; Step 5: vehicle dynamics control execution Convert path to steering angle and speed target value; Use PID algorithm to control steering system and sliding mode algorithm to control brake / drive system to track target value; Collect actual speed and steering angle, close loop adjustment control quantity to ensure accuracy; Step 6: human-machine interaction feedback and emergency takeover Feedback system status and risk information through display screen / voice feedback system; Monitor driver operation, detect voluntary operation stepwise hand over control; high risk ≥ 0.9 or failure, audible and visual alarm prompt take over, 3s no take over start backup brake; Step 7: system state monitoring and fault handling Real-time acquisition of sensor, processor, actuator, communication status; Identify fault type and level, corresponding processing; Record obstacle avoidance data for subsequent optimization; Step 8: obstacle avoidance completion and system reset Risk is reduced to When < 0.3, adjust vehicle to road speed limit, turn back to normal; Determine that the risk is removed, prompt the driver that the obstacle avoidance is completed; Reset each module to standby, wait for the next trigger.

Images