A method and storage medium for calculating the timing of automatic obstacle avoidance by large vehicles

By using a forward-facing camera to identify information about large vehicles and calculating and superimposing compensation time, the problem of insufficient safety in correcting vehicle alignment after obstacle avoidance is solved. This increases safety redundancy time and reduces hardware costs, thereby improving the safety of autonomous driving and the passenger experience.

CN123104635APending Publication Date: 2026-11-17YIBIN COWIN AUTO CO LTD
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
CN202611294860.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-25
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing technology does not add a safety margin of time when the vehicle returns to the center of the road after avoiding an obstacle, which affects the safety of autonomous driving and the passenger experience.

Method used

By using a forward-facing camera to identify the relative longitudinal distance, length, and relative speed of large vehicles in adjacent lanes, the theoretical avoidance time is calculated, and a compensation time is added on top of this to dynamically adjust the avoidance holding time. Combined with a dual return-to-center trigger mechanism of time trigger and perception confirmation, the vehicle is ensured to smoothly return to the center of the lane.

Benefits of technology

It improves the safety and passenger experience of autonomous driving, reduces hardware upgrade costs, increases safety redundancy time, enhances overtaking efficiency and driving smoothness, and adapts to different driving environments.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123104635A_ABST
    Figure CN123104635A_ABST
Patent Text Reader

Abstract

This invention provides a method and storage medium for calculating the timing of automatic large vehicle avoidance, belonging to the field of autonomous driving technology. Step 1: After the vehicle activates lane keeping function, it uses a forward-facing camera to identify the relative longitudinal distance, length, and relative speed of the large vehicle in the adjacent lane. Step 2: When the large vehicle avoidance trigger condition is met, the vehicle executes the large vehicle avoidance function, shifting laterally away from the large vehicle. Step 3: When the relative longitudinal distance between the front of the vehicle and the front of the large vehicle is detected to be 0, the theoretical avoidance time is calculated based on the length of the large vehicle and the relative speed of the two vehicles. Step 4: A compensation time is added to the theoretical avoidance time to obtain the final avoidance holding time. Step 5: The vehicle maintains the avoidance state until the final avoidance holding time is reached, then the vehicle smoothly returns to the center of the lane. This invention adds a safety redundancy time during the vehicle's return to center after avoidance, improving the safety of autonomous driving and the passenger experience.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of autonomous driving technology. Specifically, this invention relates to a method and storage medium for calculating the timing of automatic obstacle avoidance by large vehicles. Background Technology

[0002] With the rapid popularization of intelligent driving technology, L2-level lane keeping assist and large vehicle avoidance functions have gradually become the mainstream configuration of passenger vehicles. Among them, the large vehicle avoidance function can control the vehicle to slightly shift laterally away from the large vehicle when overtaking large vehicles in adjacent lanes, increasing the lateral safety distance and improving the safety and comfort of high-speed driving.

[0003] Chinese Patent 117657128A provides a collision avoidance control method and system for improving the safety of vehicle assisted driving functions. The collision avoidance control method includes: real-time acquisition of vehicle parameters, road parameters of the lane in which the vehicle is located, road condition parameters of the lane in which the vehicle is located, and parameters of other vehicles in the adjacent lane ahead; generating collision avoidance trigger conditions and lateral collision avoidance distance based on the vehicle parameters, the road parameters, and the parameters of other vehicles; if the collision avoidance trigger conditions are met, controlling the vehicle to begin lateral collision avoidance, wherein the lateral collision avoidance distance is the lateral distance the vehicle moves after initiating lateral collision avoidance; generating collision avoidance termination conditions based on the vehicle parameters, the road condition parameters, and the parameters of other vehicles, and controlling the vehicle to terminate lateral collision avoidance if the collision avoidance termination conditions are met.

[0004] Existing technology does not add a safety margin of time when the vehicle returns to the center of the road after avoiding an obstacle, which affects the safety of autonomous driving and the passenger experience. Summary of the Invention

[0005] The present invention aims to provide a method and storage medium for calculating the timing of automatic obstacle avoidance for large vehicles, so as to increase the safety redundancy time when the vehicle returns to the center of the road after obstacle avoidance, thereby improving the safety of autonomous driving and the passenger experience.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] This invention provides a method for calculating the timing of automatic obstacle avoidance by large vehicles, comprising the following steps:

[0008] Step 1: After the vehicle activates the lane keeping function, the forward-facing camera identifies the relative longitudinal distance, length, and relative speed between the vehicle and the large vehicle in the adjacent lane.

[0009] Step 2: When the large vehicle avoidance trigger condition is met, the vehicle executes the large vehicle avoidance function and shifts laterally to the side away from the large vehicle;

[0010] Step 3: When the relative longitudinal distance between the front of the vehicle and the front of the large vehicle is detected to be 0, calculate the theoretical avoidance time based on the length of the large vehicle and the relative speed of the two vehicles;

[0011] Step 4: Add compensation time to the theoretical avoidance time to obtain the final avoidance duration;

[0012] Step 5: Maintain the evasive maneuver until the final evasive maneuver duration is reached, then smoothly return the vehicle to the center of the lane.

[0013] In step one, the relative longitudinal distance is the longitudinal distance between the front of the vehicle and the rear of the truck.

[0014] In step two, the lateral offset for avoidance is dynamically calculated based on the width of the lane and the width of the large vehicle. After the offset, the distance between the vehicle and the edge line of the lane away from the large vehicle is not less than the preset safety margin.

[0015] In step three, when the front of the vehicle has passed the rear of the large vehicle but not the midpoint of the large vehicle, the relative speed and the length of the large vehicle are updated in real time by the forward-view camera, and the theoretical avoidance time is dynamically refreshed; when the front of the vehicle has passed the midpoint of the large vehicle, the parameters at the current moment are determined as the final theoretical avoidance time.

[0016] In step four, the compensation time includes the perception error compensation time, the execution delay compensation time, and the return-to-center process compensation time. The perception error compensation time is dynamically matched based on the visual perception confidence level. The execution delay compensation time is matched based on the vehicle speed. The return-to-center process compensation time is calculated based on the avoidance lateral offset and the preset return-to-center lateral acceleration.

[0017] In step five, a dual-triggering mechanism combining time-based triggering and sensory confirmation is employed:

[0018] The primary trigger condition is to achieve the duration of the final avoidance maneuver.

[0019] If the forward-facing camera detects the front of the large vehicle and the rear of the vehicle exceeds the preset safe distance from the front of the large vehicle before the timer ends, the vehicle will be automatically corrected.

[0020] When multiple large vehicles are detected in adjacent lanes in succession, and the longitudinal distance between adjacent large vehicles is less than a preset distance threshold, the avoidance time corresponding to each large vehicle is accumulated, and the vehicle is returned to center after all of them have been overtaken.

[0021] During the avoidance process, the system monitors the driver's steering intervention signal and lane line recognition status in real time. When the driver's active steering intervention is detected and the torque exceeds the takeover threshold, the system immediately exits the avoidance control and returns to the lane. When lane line recognition fails or the target of a large vehicle is lost for more than a preset time, the system adds compensation time and issues a takeover reminder. After the timeout, the system forces a smooth return to the lane.

[0022] The present invention provides a storage medium having a computer program stored thereon, which, when executed by a processor, performs the above-described method.

[0023] The technical effects of this invention are as follows:

[0024] (1) The hardware cost of the present invention is low. This solution does not require additional hardware such as side-view cameras, rear-view cameras, millimeter-wave radar, etc. It only relies on the vehicle's existing front-view camera and lane keeping controller hardware platform. All functions can be implemented through pure software algorithm deployment. It supports direct deployment to mass-produced models through the OTA cloud platform, which greatly reduces the hardware upgrade cost and function implementation cycle. It can quickly cover existing models with L2-level lane keeping function that have been launched.

[0025] (2) The avoidance time calculation of this invention is accurate, taking into account both overtaking efficiency and driving safety. It adopts a segmented parameter update mechanism. During the stage before the front of the vehicle exceeds the midpoint of the vehicle body, the length and relative speed parameters of the vehicle body are updated in real time to dynamically correct the theoretical avoidance time. After the front of the vehicle crosses the midpoint of the vehicle body, the calculation parameters are locked, which ensures the timeliness of the calculation in the early stage of overtaking and avoids the time error caused by parameter fluctuations in the second half of the process. It sets three independent compensation components: perception error, execution delay, and return process. Among them, the perception error compensation is dynamically matched with the visual perception confidence, and the execution delay compensation is dynamically matched with the visual perception confidence. The system features graded vehicle speed adaptation and precise calculation of offset and return-to-center acceleration to compensate for the return-to-center process. It comprehensively covers error sources such as visual recognition error, hardware response latency, and return-to-center action time, significantly improving the calculation accuracy of the final avoidance holding time. It adopts a dual return-to-center triggering mechanism that combines time triggering and perception confirmation. The calculated final avoidance time is used as the core triggering condition to ensure safety redundancy. At the same time, it supports early triggering of return-to-center when the rear of a vehicle exceeds the safe distance of the front of a large vehicle, avoiding the vehicle from deviating from the center of the lane for a long time for no reason, thus improving driving smoothness and traffic efficiency.

[0026] (3) This invention improves the safety of large vehicle avoidance. The lateral offset of the avoidance is dynamically calculated based on the width of the lane and the width of the large vehicle, while constraining the safety boundaries on both sides to ensure the lateral safety distance during overtaking; it achieves a balance between safe avoidance and compliant driving within the limited lane space and can be adapted to lane conditions of different widths.

[0027] (4) This invention has multiple safety redundancy mechanisms and sets strict avoidance trigger access conditions. Avoidance is triggered only when the lane is centered and stable, the width of the large vehicle meets the standard, there is a stable overtaking trend, the road is straight, and there are no large vehicles in close proximity. This reduces the risk of false triggering from the source. It supports a driver priority takeover mechanism. During the avoidance process, the driver's steering torque is detected in real time. When the torque exceeds the takeover threshold, the automatic avoidance control is immediately exited and the steering authority is returned to ensure the highest priority of the driver's operation. It has a fallback logic for handling abnormal perception. When the lane line recognition fails or the target of the large vehicle is lost for more than a preset time, an additional compensation time is added and a takeover reminder is issued. After the timeout, the vehicle is forced to return to center smoothly to avoid the vehicle from continuously deviating from the lane after the perception fails. It sets two-level boundary protection. When the relative vehicle speed is too low or the overtaking trend disappears, the avoidance is immediately terminated and a return to center reminder is issued. At the same time, a maximum avoidance time limit is set. Under extreme conditions, the vehicle is forced to return to center to eliminate the safety hazard of the vehicle deviating from the center of the lane for a long time.

[0028] (5) The present invention is highly adaptable to the environment. It comprehensively evaluates the confidence of visual perception from three dimensions: image exposure, wiper gear, and completeness of vehicle outline. The lowest value of the three is used as the final confidence judgment result. Based on the confidence level, the perception error compensation time is dynamically matched. It can adapt to different driving environments such as backlight, underexposure, and rainy weather. When the perception conditions are reduced, the compensation time is increased to ensure avoidance safety. When the perception conditions are good, the compensation time is controlled to improve traffic efficiency, thus achieving a dynamic balance between safety and efficiency. Attached Figure Description

[0029] This manual includes the following figures, which illustrate the following:

[0030] Figure 1 This is a flowchart illustrating a method and storage medium for calculating the timing of automatic obstacle avoidance by large vehicles according to the present invention. Detailed Implementation

[0031] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings, in order to help those skilled in the art to have a more complete, accurate and in-depth understanding of the inventive concept and technical solution of the present invention, and to facilitate its implementation.

[0032] This invention provides a method for calculating the timing of automatic obstacle avoidance by large vehicles, comprising the following steps:

[0033] Step 1: After the vehicle activates the lane keeping function, the forward-facing camera identifies the relative longitudinal distance, length, and relative speed between the vehicle and the large vehicle in the adjacent lane.

[0034] Step 2: When the large vehicle avoidance trigger condition is met, the vehicle executes the large vehicle avoidance function and shifts laterally to the side away from the large vehicle;

[0035] Step 3: When the relative longitudinal distance between the front of the vehicle and the front of the large vehicle is detected to be 0, calculate the theoretical avoidance time based on the length of the large vehicle and the relative speed of the two vehicles;

[0036] Step 4: Add compensation time to the theoretical avoidance time to obtain the final avoidance duration;

[0037] Step 5: Maintain the evasive maneuver until the final evasive maneuver duration is reached, then smoothly return the vehicle to the center of the lane.

[0038] In step one, the relative longitudinal distance is the longitudinal distance between the front of the vehicle and the rear of the truck.

[0039] In step two, the lateral offset for avoidance is dynamically calculated based on the width of the lane and the width of the large vehicle. After the offset, the distance between the vehicle and the edge line of the lane away from the large vehicle is not less than the preset safety margin.

[0040] In step three, when the front of the vehicle has passed the rear of the large vehicle but not the midpoint of the large vehicle, the relative speed and the length of the large vehicle are updated in real time by the forward-view camera, and the theoretical avoidance time is dynamically refreshed; when the front of the vehicle has passed the midpoint of the large vehicle, the parameters at the current moment are determined as the final theoretical avoidance time.

[0041] In step four, the compensation time includes the perception error compensation time, the execution delay compensation time, and the return-to-center process compensation time. The perception error compensation time is dynamically matched based on the visual perception confidence level. The execution delay compensation time is matched based on the vehicle speed. The return-to-center process compensation time is calculated based on the avoidance lateral offset and the preset return-to-center lateral acceleration.

[0042] In step five, a dual-triggering mechanism combining time-based triggering and sensory confirmation is employed:

[0043] The primary trigger condition is to achieve the duration of the final avoidance maneuver.

[0044] If the forward-facing camera detects the front of the large vehicle and the rear of the vehicle exceeds the preset safe distance from the front of the large vehicle before the timer ends, the vehicle will be automatically corrected.

[0045] When multiple large vehicles are detected in adjacent lanes in succession, and the longitudinal distance between adjacent large vehicles is less than a preset distance threshold, the avoidance time corresponding to each large vehicle is accumulated, and the vehicle is returned to center after all of them have been overtaken.

[0046] During the avoidance process, the system monitors the driver's steering intervention signal and lane line recognition status in real time. When the driver's active steering intervention is detected and the torque exceeds the takeover threshold, the system immediately exits the avoidance control and returns to the lane. When lane line recognition fails or the target of a large vehicle is lost for more than a preset time, the system adds compensation time and issues a takeover reminder. After the timeout, the system forces a smooth return to the lane.

[0047] The present invention provides a storage medium having a computer program stored thereon, which, when executed by a processor, performs the above-described method.

[0048] The present invention will now be described in detail.

[0049] This invention provides a method for calculating the timing of automatic lane avoidance by large vehicles, applicable to vehicles equipped with cameras and Level 2 lane keeping assist functionality. This method requires no additional hardware such as side-view cameras, rear-view cameras, or millimeter-wave radar; it is implemented entirely through software algorithms deployed within existing lane keeping controllers and can be directly deployed to mass-produced vehicles via OTA cloud platforms.

[0050] After the vehicle is powered on and the lane-keeping function is activated by the driver, the system enters standby mode, simultaneously collecting two types of input data and performing validity checks and preprocessing. The first type is visual perception data, output by the forward-facing camera, specifically including: the longitudinal relative distance (s1) between the front of the vehicle and the rear of the larger vehicle in adjacent lanes: defined as the longitudinal distance between the front of the vehicle and the rear of the larger vehicle; s1 is positive when the vehicle is behind the larger vehicle and negative after the vehicle has passed its rear; the length of the larger vehicle (x1): calculated by visually recognizing the feature points of the front and rear of the larger vehicle and combining this with a monocular ranging model; the relative speed (v1) between the vehicle and the larger vehicle: defined as the vehicle speed minus the larger vehicle speed; a positive value indicates that the vehicle is moving forward relative to the larger vehicle and has an overtaking tendency; the position of the left and right lane lines; and the width of the current lane. The second type is vehicle status data, acquired through the vehicle's CAN bus and the vehicle's inertial measurement unit (IMU), specifically including: the vehicle's actual speed v0, the driver's steering torque signal, and the windshield wiper setting. After all raw data enters the controller, it undergoes preprocessing: Kalman filtering is used to smooth s1, x1, and v1, eliminating abrupt noise and improving data stability; simultaneously, based on three dimensions—image exposure, wiper mode, and vehicle outline completeness—the current perception confidence is divided into high, medium, and low levels for dynamic matching of subsequent compensation parameters. Specifically, image exposure is determined based on the image's average grayscale value and automatic exposure gain. When the image's average grayscale value is in the range of 80~180 and the automatic exposure gain is ≤1, it is judged as high confidence; when the image's average grayscale value is in the range of 40~80 (underexposed) or 180~220 (overexposed) and the automatic exposure gain is 1~3 times, it is judged as medium confidence; when the image's average grayscale value is <40 or >220 and the automatic exposure gain is >3 times, it is judged as low confidence. The confidence level is determined as follows: high confidence when the wipers are in the off position; medium confidence when the wipers are in the intermittent or low-speed position; and low confidence when the wipers are in the high-speed position. The completeness of the vehicle's outline is determined by the detection rate of key outline points on the vehicle body; a detection rate of 90% or higher is considered high confidence; a detection rate of 70% or higher but less than 90% is considered medium confidence; and a detection rate of less than 70% is considered low confidence. The lowest confidence level among the three dimensions is used as the final confidence level.

[0051] The system continuously checks whether all of the following triggering conditions are met simultaneously. If all conditions are met, the system enters the avoidance execution state. If any condition is not met, the system maintains its position in the center of the normal lane to avoid false triggering. These conditions include: determining that the vehicle is in a stable, centered position in the lane based on collected image data; identifying a large vehicle with a width greater than a preset threshold (2.4m) in an adjacent lane in the same direction; the relative speed being greater than the minimum overtaking relative speed calibration value, which is 5km / h in this embodiment, confirming that the vehicle has a stable overtaking trend and excluding parallel driving at the same speed; the current road curvature radius calculated from the lane line curvature being greater than a preset threshold (greater than 500m), indicating a straight road with sufficient lateral avoidance space and no risk of crossing the line; and no two consecutive large vehicles with a small longitudinal distance (less than 20m) detected ahead in adjacent lanes. If multiple consecutive vehicles are detected, the system enters the consecutive multiple vehicle avoidance branch logic.

[0052] After entering the avoidance execution state, the vehicle shifts laterally a preset distance away from the larger vehicle, while continuously monitoring the relative longitudinal distance s1. When the relative longitudinal distance is 0, the theoretical avoidance time is calculated. Specifically, when the front of the vehicle has not passed the midpoint of the larger vehicle, the forward-looking camera collects the vehicle length and relative speed information in real time, and calculates and updates the theoretical avoidance time in real time; when the front of the vehicle passes the midpoint of the larger vehicle, the system locks the vehicle length and relative speed of the current frame as the final calculation parameters, and the theoretical avoidance time is no longer updated.

[0053] A compensation time is added to the theoretical avoidance time. This compensation time includes three independently calibrable compensation components, each corresponding to a different source of error. The compensation components include the sensing error compensation t. 31 Execution delay compensation t 32 Compensation t during the correction process 33 The perception error compensation is used to offset the inherent visual error in large vehicle length recognition, and its value is dynamically determined based on the vehicle model and perception confidence level. In this embodiment, the calibration values ​​are: 0.15s for high confidence, 0.35s for medium confidence, and 0.55s for low confidence. The execution delay compensation is used to cover the hardware delay of controller calculation and steering motor response, and its value is matched to the vehicle speed v0. In this embodiment, the calibration values ​​are: 0.1s for 60km / h≤v0<90km / h, 0.2s for 90km / h≤v0<120km / h, and 0.25s for v0≥120km / h. The return-to-center process compensation is used to offset the travel time of the vehicle's return-to-center action, ensuring that the vehicle returns to the center of the lane exactly when the timing ends; the time required for the entire return-to-center process is calculated based on the current avoidance lateral offset d and the preset maximum return-to-center lateral acceleration a (calibrated to 0.15g, where g is the acceleration due to gravity), using the following formula:

[0054]

[0055] The formula for calculating the final avoidance time and the total time t2 of the final avoidance with boundary constraints is: t2 = t1 + t 31 +t 32 +t 33

[0056] At the same time, a two-level boundary protection mechanism is set to avoid system abnormalities under extreme conditions: if the real-time relative vehicle speed v1 < 2km / h, it is determined that the overtaking trend has disappeared, the avoidance timer is immediately terminated, a smooth return to center is triggered, and a driver takeover reminder is issued through the instrument panel; the maximum avoidance time limit is set to 15s. If the return to center condition is not met after 15s, a smooth return to center is forcibly triggered and a driver takeover reminder is issued to avoid system abnormalities that cause the vehicle to deviate from the center of the lane for a long time.

[0057] During the avoidance process, the lateral offset of the vehicle is dynamically calculated based on the width of the lane. Specifically, after the offset, the lateral distance between the edge of the vehicle body closer to the large vehicle and the large vehicle body is not less than 0.3m, and the distance between the wheels of the vehicle on the side farther from the large vehicle and the lane edge line is not less than 0.3m. In this embodiment, the typical offset under a standard 3.75m lane is 0.3~0.5m, ensuring a safe avoidance distance while avoiding the risk of crossing the lane line. Simultaneously, the system monitors the entire process to ensure human-machine co-driving safety. Specifically, it collects the driver's steering torque in real time. When the torque exceeds the preset takeover threshold (calibrated to 3N·m), it determines that the driver has actively intervened in the steering, immediately exits automatic avoidance control, returns steering authority to the driver, and the vehicle naturally returns to center with the driver's input. The system continuously monitors the lane line recognition status and the large vehicle target tracking status. If the lane line or large vehicle target is lost, the automatic avoidance control is immediately exited.

[0058] When the obstacle avoidance timer reaches its final duration, the controller sends a return-to-center command to the steering system. The vehicle smoothly returns to the center of the lane with the calibrated lateral acceleration, and the obstacle avoidance function ends. If, before the timer expires, the forward-facing camera detects the front of a large vehicle and calculates that the front of the vehicle has exceeded the safe distance of 5 meters from the front of the large vehicle, the system can terminate the timer early and trigger the return-to-center function to avoid unnecessary prolonged drift.

[0059] The beneficial effects of the present invention are described in detail below.

[0060] The hardware cost of this invention is low. This solution does not require additional hardware such as side-view cameras, rear-view cameras, or millimeter-wave radar. It only relies on the vehicle's existing front-view camera and lane-keeping controller hardware platform. All functions can be implemented through pure software algorithm deployment. It supports direct deployment to mass-produced models via OTA cloud platform, which greatly reduces hardware upgrade costs and function implementation cycle. It can quickly cover existing models with L2-level lane-keeping function that are already on the market.

[0061] This invention provides precise calculation of avoidance time, balancing overtaking efficiency and driving safety. It employs a segmented parameter update mechanism: while the vehicle's front end is still within the midpoint of the larger vehicle's body, the length and relative speed parameters of the larger vehicle are updated in real time to dynamically correct the theoretical avoidance time. Once the front end of the vehicle has passed the midpoint, the calculation parameters are locked, ensuring the timeliness of the calculation in the early stages of overtaking while avoiding time errors caused by parameter fluctuations in the latter half. Three independent compensation components are set: perception error, execution delay, and return-to-center process. The perception error compensation dynamically matches the visual perception confidence level, and the execution delay compensation varies with the vehicle's position. The system features speed-level adaptation and compensation during the return-to-center process. It combines offset and return-to-center acceleration for precise calculation, comprehensively covering error sources such as visual recognition error, hardware response latency, and return-to-center action time, significantly improving the calculation accuracy of the final avoidance holding time. It adopts a dual return-to-center triggering mechanism that combines time triggering and perception confirmation. The calculated final avoidance time is used as the core triggering condition to ensure safety redundancy. At the same time, it supports early triggering of return-to-center when the rear of a vehicle exceeds the safe distance of the front of a large vehicle, avoiding meaningless long-term deviation of the vehicle from the center of the lane, and improving driving smoothness and traffic efficiency.

[0062] This invention improves the safety of large vehicle avoidance. The lateral offset for avoidance is dynamically calculated based on the width of the lane and the width of the large vehicle, while simultaneously constraining the safety boundaries on both sides to ensure a safe lateral distance during overtaking. It achieves a balance between safe avoidance and compliant driving within limited lane space and can be adapted to lane conditions of different widths.

[0063] This invention features multiple safety redundancy mechanisms and sets strict avoidance trigger conditions. Avoidance is only triggered when conditions such as stable lane centering, adequate vehicle width, a stable overtaking trend, a straight road, and no closely spaced large vehicles are simultaneously met, reducing the risk of false triggering from the outset. It supports a driver-priority takeover mechanism, monitoring the driver's steering torque in real time during avoidance. When the torque exceeds the takeover threshold, automatic avoidance control is immediately disengaged, and steering authority is returned, ensuring the highest priority for driver operation. It also has a fallback logic for handling abnormal perception. When lane line recognition fails or the target vehicle is lost for more than a preset time, additional compensation time is added and a takeover reminder is issued. After the timeout, a forced and smooth return to center is achieved, preventing the vehicle from continuously deviating from the lane after perception failure. Two levels of boundary protection are set. When the relative speed is too low or the overtaking trend disappears, avoidance is immediately terminated and a return-to-center reminder is issued. A maximum avoidance duration limit is also set, forcing a return to center under extreme conditions to eliminate the safety hazard of the vehicle deviating from the lane center for extended periods.

[0064] This invention is highly adaptable to the environment. It comprehensively evaluates the confidence level of visual perception from three dimensions: image exposure, wiper mode, and completeness of vehicle outline. The lowest value of the three is used as the final confidence level. Based on the confidence level, the invention dynamically matches the perception error compensation time, which can adapt to different driving environments such as backlight, underexposure, and rainy weather. When the perception conditions are reduced, the compensation time is increased to ensure avoidance safety. When the perception conditions are good, the compensation time is controlled to improve traffic efficiency, thus achieving a dynamic balance between safety and efficiency.

[0065] The present invention has been described above by way of example with reference to the accompanying drawings. Obviously, the specific implementation of the present invention is not limited to the above-described manner. Any non-substantial improvements made using the inventive concept and technical solution; or the direct application of the inventive concept and technical solution to other situations without modification, are all within the protection scope of the present invention.

Claims

1. A method for calculating the timing of automatic obstacle avoidance by large vehicles, characterized in that, Includes the following steps: Step 1: After the vehicle activates the lane keeping function, the forward-facing camera identifies the relative longitudinal distance, length, and relative speed between the vehicle and the large vehicle in the adjacent lane. Step 2: When the large vehicle avoidance trigger condition is met, the vehicle executes the large vehicle avoidance function and shifts laterally to the side away from the large vehicle; Step 3: When the relative longitudinal distance between the front of the vehicle and the front of the large vehicle is detected to be 0, calculate the theoretical avoidance time based on the length of the large vehicle and the relative speed of the two vehicles. Step 4: Add compensation time to the theoretical avoidance time to obtain the final avoidance duration; Step 5: Maintain the evasive maneuver until the final evasive maneuver duration is reached, then smoothly return the vehicle to the center of the lane.

2. The method for calculating the timing of automatic obstacle avoidance by large vehicles as described in claim 1, characterized in that: In step one, the relative longitudinal distance is the longitudinal distance between the front of the vehicle and the rear of the truck.

3. The method for calculating the timing of automatic obstacle avoidance by large vehicles as described in claim 1, characterized in that: In step two, the lateral offset for avoidance is dynamically calculated based on the width of the lane and the width of the large vehicle. After the offset, the distance between the vehicle and the edge line of the lane away from the large vehicle is not less than the preset safety margin.

4. The method for calculating the timing of automatic obstacle avoidance by large vehicles as described in claim 1, characterized in that: In step three, when the front of the vehicle has passed the rear of the large vehicle but not the midpoint of the large vehicle, the relative speed and the length of the large vehicle are updated in real time by the forward-view camera, and the theoretical avoidance time is dynamically refreshed; when the front of the vehicle has passed the midpoint of the large vehicle, the parameters at the current moment are determined as the final theoretical avoidance time.

5. The method for calculating the timing of automatic obstacle avoidance by large vehicles as described in claim 1, characterized in that: In step four, the compensation time includes the perception error compensation time, the execution delay compensation time, and the return-to-center process compensation time. The perception error compensation time is dynamically matched based on the visual perception confidence level. The execution delay compensation time is matched based on the vehicle speed. The return-to-center process compensation time is calculated based on the avoidance lateral offset and the preset return-to-center lateral acceleration.

6. The method for calculating the timing of automatic obstacle avoidance by large vehicles as described in claim 1, characterized in that: In step five, a dual-triggering mechanism combining time-based triggering and sensory confirmation is employed: The primary trigger condition is to achieve the duration of the final avoidance maneuver. If the forward-facing camera detects the front of the large vehicle and the rear of the vehicle exceeds the preset safe distance from the front of the large vehicle before the timer ends, the vehicle will be automatically corrected.

7. The method for calculating the timing of automatic obstacle avoidance by large vehicles as described in claim 1, characterized in that: When multiple large vehicles are detected in adjacent lanes in succession, and the longitudinal distance between adjacent large vehicles is less than a preset distance threshold, the avoidance time corresponding to each large vehicle is accumulated, and the vehicle is returned to center after all of them have been overtaken.

8. The method for calculating the timing of automatic obstacle avoidance by large vehicles as described in claim 1, characterized in that: During the avoidance process, the system monitors the driver's steering intervention signal and lane line recognition status in real time; when the system detects the driver's active steering intervention and the torque exceeds the takeover threshold, it immediately exits the avoidance control and returns to the lane. When lane line recognition fails or the target of a large vehicle is lost for more than a preset time, additional compensation time is added and a takeover reminder is issued. After the timeout, the vehicle is forced to return to center smoothly.

9. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it performs the method as described in any one of claims 1 to 8.

Citation Information

Patent Citations

  • Avoidance control method and system for improving safety of auxiliary driving function of vehicle

    CN117657128A