Lane changing and narrow parking space reciprocating oblique column control method for IGV

Abstract

The application discloses a lane changing and narrow parking space reciprocating oblique column control method of IGV, and relates to the technical field of automatic driving of automatic wharf of port. The method comprises the following steps: scene perception and operation type identification; when it is determined that the scene is a narrow parking space column control scene, reciprocating oblique column control is performed, including double-condition quantification access determination, dynamic safety approach distance setting, reciprocating cycle of forward and backward oblique column to the target direction, calculation of lateral deviation and longitudinal cumulative displacement, and column completion determination; when it is determined that the scene is a lane changing and parking scene, lane changing and parking control is performed, including empty parking space identification, lane changing feasibility determination, smooth lane changing path generation, and reciprocating oblique parking along the path. The application also supports networking main control driven by communication state perception and dual-mode adaptive switching of network redundancy, and realizes safe, accurate and efficient control of IGV in the dense parking space scene of port.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving technology, and in particular to a lane changing and narrow parking space reciprocating diagonal lane exit control method applicable to IGV. Background Technology

[0002] Currently, mainstream IGV motion control solutions in the industry are difficult to adapt to the extreme conditions of dense parking spaces. Conventional path-tracking steering control relies on a fixed turning radius to complete lane changes. The turning radius of the vehicle is constrained by its own wheelbase, and the required turning space is usually 1.2 to 1.5 times the total length of the vehicle. When the front and rear clearance of the parking space is insufficient, the vehicle is very likely to collide with surrounding obstacles during steering. At the same time, this type of solution only supports single continuous steering and does not have the ability to make small back-and-forth attitude adjustments, making it impossible to drive out of narrow parking spaces. Lane-changing solutions that rely on fixed lane lines and dedicated buffer areas depend on complete structured lane markings and sufficient lane-changing space. However, lane lines in port yards are often obstructed by containers, and there are generally obstacles around the target parking space with no extra buffer distance. A smooth trajectory planned in one go will interfere with surrounding facilities, and the driving trajectory cannot be dynamically adjusted according to real-time obstacle information throughout the process, resulting in weak environmental adaptability.

[0003] The narrow-lane parking exit scheme, which relies on reversing the historical parking entry path, depends entirely on complete trajectory storage data. Once the vehicle is moved by other equipment or its initial parking posture deviates, the original path data becomes invalid. This scheme can only reverse along the original path and lacks autonomous posture correction and multiple fine-tuning logic. When the available longitudinal space of the parking space is insufficient to support a single diagonal movement, the vehicle will get stuck. Furthermore, it can only solve the single parking exit scenario and cannot accommodate lane changing and parking needs. Another type of positioning and parking method combines a single diagonal movement with a single reversing movement. This is only suitable for open T-junctions without obstructions. It relies on a single-wheel diagonal reversing movement to complete the positioning and lacks multi-round iterative approach control logic. The displacement that can be adjusted in a single action is limited, and it cannot gradually accumulate the exit displacement in the narrow space squeezed by obstacles in front and behind. This scheme only supports entry and parking operations and lacks the ability to control exiting restricted parking spaces, nor has it established a unified dynamic safety braking logic.

[0004] In summary, existing control schemes generally suffer from insufficient spatial adaptability. Traditional steering mechanisms require large turning spaces, and single-stage diagonal adjustments lack a cyclic stepping mechanism, making vehicles prone to jamming when longitudinal clearance is insufficient. All adjustment actions are executed only once, lacking convergent iterative control logic and corresponding safety constraints such as cycle termination, in-situ oscillation recognition, and timeout shutdown. Obstacle perception and vehicle braking control are independent, with safe stopping distances relying on fixed empirical values, failing to adapt to vehicle speed and road surface adhesion. This results in wasted operating space at low speeds and insufficient braking safety margin at high speeds. Furthermore, existing solutions do not consider autonomous operation capabilities after communication link interruptions. Once vehicle-road cooperative communication is lost, vehicles cannot independently complete lane changes and exit maneuvers, directly disrupting continuous dock operations. Based on these numerous shortcomings of existing technologies, the industry urgently needs an IGV control method that can simultaneously cover both lane changing and exiting confined spaces, incorporates iterative convergent control for diagonal driving, adaptive dynamic safety braking, and supports redundant autonomous operation with both network connectivity and offline connectivity. Summary of the Invention

[0005] In view of this, the present invention provides a lane changing and narrow parking space reciprocating oblique exit control method for IGV, which aims to solve the technical problems in the prior art that lane changing cannot adapt to dense parking space scenarios, narrow parking spaces cannot be safely exited, there is a lack of multi-step reciprocating convergent adjustment mechanism, and the inability to operate when communication is interrupted.

[0006] To achieve the above objectives, the present invention provides the following technical solution.

[0007] This invention provides a lane changing and narrow parking space reciprocating oblique exit control method for IGV, including a communication status perception and operation mode selection step S0, a scene perception and operation type identification step S1, a narrow parking space exit control step S2, and a lane changing and entry control step S3.

[0008] In the communication status awareness and operation mode selection step S0, the IGV is configured with dual communication terminals and multiple network interfaces to monitor the channel quality of each communication link in real time, and determines the communication quality level based on signal strength and signal-to-noise ratio. When the quality of all communication links is lower than a preset threshold for a continuous period of time, or when heartbeat packets with the fleet management system are lost more than a preset number of times, or when the primary communication network is disconnected and redundant links are unavailable, the vehicle is determined to be in a network outage state, and automatically switches to the network outage redundancy mode. If the communication quality is good, the vehicle operates in the network main control mode.

[0009] In the scene perception and operation type identification step S1, the IGV's pose information, obstacle distance information, and target lane status information are acquired to determine whether the current operation scenario is a narrow parking space exit scenario or a lane change into a parking space scenario. The pose information includes the vehicle's coordinates, heading angle, front axle equivalent rotation angle, and rear axle equivalent rotation angle in the geodetic coordinate system. The obstacle distance information includes the distance to the nearest obstacle in front and the distance to the nearest obstacle behind. The target lane status information includes the centerline equation of the target lane, the position and size of the available parking spaces in the target lane, and the distribution of obstacles in the target lane.

[0010] When the scenario is determined to be a narrow parking space exit scenario, the reciprocating diagonal exit control process is executed.

[0011] First, a feasibility assessment for exiting the lane is performed (step S21), based on a dual-condition access determination using the distances to obstacles ahead and behind. This invention employs a dual-condition access logic: firstly, a sufficient longitudinal space condition, requiring the total available space ahead and behind to be greater than the sum of the vehicle's total length and safety redundancy, ensuring the vehicle has basic maneuverability within the parking space; secondly, an insufficient forward space trigger condition, requiring the forward space to be less than the minimum safe distance threshold required to complete a full lane change in diagonal mode. This threshold is determined by a combination of lane width, vehicle dimensions, maximum diagonal deviation angle, and a safety factor. Only when both conditions are simultaneously met is the reciprocating diagonal exit mode triggered.

[0012] If the longitudinal space requirement is not met, a pre-adjustment to the intermediate position is performed: a yaw moment is generated by differential steering of the front and rear axles, causing the vehicle to make a small-scale attitude adjustment within the parking space to redistribute the front and rear space margins, and the dual-condition judgment is performed again. If the longitudinal space requirement is still not met after the pre-adjustment, it is determined that de-queueing is not feasible, and the safety protection mechanism is triggered.

[0013] After entering the reciprocating diagonal de-queue mode, during each diagonal movement, the present invention calculates the dynamic safe approach distance in real time (step S22). This distance is calculated in real time based on the current vehicle speed, system response delay, and maximum braking deceleration, and can be corrected according to the road surface adhesion coefficient to ensure the reliability of the braking distance under different road surface conditions. When the distance to the obstacle detected in real time is less than or equal to the dynamic safe approach distance, a braking stop command is immediately triggered.

[0014] The vehicle then switches to diagonal driving mode and moves diagonally forward in the target direction (step S23). This invention uses an omnidirectional diagonal kinematic model to control the vehicle, achieving overall vehicle translation through unidirectional deflection of the front and rear axles, bringing the vehicle's center of gravity sideslip angle close to zero and avoiding yaw motion. During the forward diagonal movement, the distance to obstacles ahead is detected in real time, and the vehicle brakes and stops when a dynamic safe approach distance is reached, recording the distance traveled.

[0015] Next, the vehicle switches to a backward diagonal movement towards the target direction (step S24), where the target direction angle for the backward diagonal movement is the opposite of the forward diagonal direction angle. During the backward diagonal movement, the invention uses an attitude deviation control law to adjust the difference in the front and rear axle rotation angles to generate a corrective yaw moment, adaptively converging the vehicle's heading deviation to gradually converge the deviation between the vehicle's heading and the target lane centerline to within a preset attitude deviation threshold. The distance to obstacles behind the vehicle is detected in real time during the backward diagonal movement, and braking stops and the backward diagonal distance is recorded when a dynamic safe approach distance is reached.

[0016] After completing one full cycle of "forward diagonal movement + backward diagonal movement" (step S25), calculate the lateral deviation of the current vehicle relative to the centerline of the target lane and the cumulative longitudinal displacement of the vehicle that has moved out of its original parking space. The cumulative longitudinal displacement is the sum of the longitudinal components of the forward diagonal movement distance and the backward diagonal movement distance in each cycle.

[0017] Simultaneously, single-cycle displacement monitoring is performed: if the actual longitudinal displacement increment in a certain cycle is less than the preset minimum effective displacement threshold, it is determined that the vehicle has fallen into a stationary oscillation state, the reciprocating cycle is immediately terminated, and the safety protection mechanism is triggered.

[0018] The determination of successful exit from the lane (step S26) requires the simultaneous fulfillment of three conditions: the lateral deviation is less than a preset threshold, the longitudinal cumulative displacement reaches the minimum displacement required for exiting, and the current heading deviation is less than a preset threshold. When all three conditions are met, it is determined that the vehicle has completely exited the original parking space and successfully entered the target lane, thus exiting the reciprocating diagonal driving mode.

[0019] If the de-queue is not completed and the current cycle count is less than the maximum allowed cycle count (step S27), return to continue the next round of repetition; if the maximum cycle count has been reached, trigger the safety protection mechanism. The maximum cycle count is dynamically set according to the current parking space size.

[0020] When the scenario is determined to be a lane change into a parking space, the lane change into a parking space control procedure is executed.

[0021] First, identify available parking spaces in the target lane (step S31). Perform a gridded scan along the lane direction within the target lane area, count the obstacle point cloud density in each grid, identify continuous grid areas with point cloud density below a preset threshold as available parking spaces, and calculate the length of the available parking space.

[0022] Then, the feasibility of lane changing is determined (step S32). When the length of the available parking space is greater than the sum of the total length of the vehicle and the safe clearance, the collision risk index of the lane changing path is further calculated. When the collision risk index is lower than the preset threshold, it is determined that there are no obstacles in the lane changing path, and the lane changing process is triggered.

[0023] Then, a smooth lane change path is generated (step S33). Taking the vehicle's current position as the lane change starting point and the center of the empty parking space as the lane change ending point, a fifth-order polynomial is used to generate a smooth lane change path. The boundary constraints satisfy the continuity of the starting and ending point positions, heading angles, and curvatures, and the maximum curvature of the path is less than the maximum achievable curvature of the vehicle's steering system.

[0024] The generated fifth-order polynomial path is discretized into path points, and the vehicle tracks these path points in an oblique driving mode (step S34). Within each control cycle, the lateral and heading deviations from the vehicle's current position to the path are calculated. The desired front and rear axle rotation angles are calculated using an oblique kinematic model, enabling the vehicle to travel obliquely along the path. The target speed is set to low, and the distance to obstacles ahead is detected in real time. The same dynamic safe approach distance as in the narrow parking space exit scenario is used for safe braking determination.

[0025] Switch to diagonal driving in the target direction (step S35), adjust the vehicle's heading to be consistent with the center line of the target lane through the attitude convergence control law, detect the distance to the obstacle behind in real time during the diagonal driving process, and stop diagonal driving when the dynamic safe approach distance is reached.

[0026] Calculate the lateral deviation of the current vehicle relative to the center line of the target lane (step S36). If the lateral deviation is less than the preset threshold, it is determined that the vehicle has completely entered the target parking space. If the threshold is not met and the number of cycles has not reached the upper limit, repeat the forward and backward diagonal operations.

[0027] When the positioning condition is met or the maximum number of cycles is reached, exit the diagonal driving mode (step S37). If the positioning is successful, resume normal driving mode; if the positioning fails to occur within the time limit, trigger safety protection and report an anomaly.

[0028] When in network-off redundancy mode, the vehicle loads locally cached maps and lane geometry information, and the perception system switches to pure vehicle-mounted perception mode. In tight parking space exit control, target lane availability is confirmed through active LiDAR scanning, scanning the boundaries of both sides of the target lane, clustering obstacle point clouds within the target area, identifying the start and end boundaries of the available space, and calculating its usable length along the lane direction. In lane-to-lane change and entry control, a gridded scan is performed along the lane direction within the target lane area to identify available parking spaces. Dynamic safe approach distance calculation uses conservative parameters, and the maximum number of loops is reduced in network-off mode. When the communication link is detected to be restored, a smoothing filter is used to transition the control commands.

[0029] Compared with the prior art, the present invention has the following significant advantages: This invention simultaneously covers two core scenarios: two-way lane changing into a parking space and reciprocating diagonal exiting a parking space in confined spaces. By using scenario perception and operation type identification to determine the working condition, the two scenarios are uniformly mapped to the core control logic of "forward diagonal movement + backward diagonal movement + cyclic determination". The safe distance calculation model, kinematic control law and mode switching mechanism are shared, which reduces system complexity and improves engineering feasibility.

[0030] The dual-condition access logic proposed in this invention, consisting of "sufficient longitudinal space condition + insufficient forward space trigger condition," is calculated based on vehicle kinematic parameters, lane geometric parameters, and safety redundancy coefficients. It possesses clear physical meaning and repeatability. The dynamic safe approach distance incorporates system response delay, braking deceleration, and parking safety clearance into a unified quantitative model, enabling real-time adaptive adjustment of the safe distance according to vehicle speed. This adaptability to different road conditions simultaneously maximizes the balance between space utilization efficiency and safety.

[0031] This invention establishes a complete kinematic model of IGV in slant mode. By using a differential steering angle control strategy, it achieves independent control of slant translation and heading correction, so that the vehicle's attitude always converges to the target lane direction during the reciprocating cycle, ensuring the repeatability and predictability of multiple cycles and avoiding the collision risk caused by attitude divergence.

[0032] This invention establishes a mathematically convergent reciprocating oblique loop control logic: the lateral deviation successively approaches the center line of the target lane, the longitudinal cumulative displacement monotonically increases with the number of loops, and the maximum number of loops is dynamically set to ensure that the loop terminates within a finite number of steps, thus completely avoiding the theoretical risk of "infinite loop".

[0033] This invention employs a combined motion strategy of "forward oblique movement in the target direction + backward oblique movement in the target direction," using an oblique movement mode throughout the entire process. The vehicle moves horizontally as a whole without producing significant yaw motion, fundamentally eliminating the turning space required for traditional front axle steering and the fishtailing effect of all-wheel steering. Even in extreme conditions where the longitudinal safety distance is only a few centimeters, it can still safely exit the formation by gradually accumulating displacement through single, small steps.

[0034] In lane-changing and parking scenarios, a fifth-order polynomial is used to generate a smooth path that satisfies the triple constraints of position, heading angle, and curvature continuity. The maximum curvature constraint ensures the drivability of the path. An available parking space recognition method based on point cloud density statistics achieves accurate parking space detection.

[0035] This invention integrates four sensing channels: combined navigation, lidar, ultrasonic sensors, and vehicle-to-infrastructure (V2I) communication, enabling redundant detection and cross-verification of pose and obstacle information. Simultaneously, it constructs a three-tiered safety protection mechanism: dynamic safe distance real-time braking, single-cycle displacement monitoring, and maximum cycle count protection, progressively ensuring the system can safely stop under any abnormal operating conditions.

[0036] The entire process supports bidirectional adaptation. By reversing the target direction angle, the operation of leaving the queue / changing lanes to the left or right can be achieved without modifying the core control algorithm. It can flexibly adapt to the lane changing and leaving lane requirements of different port yard layouts and different operating directions, significantly reducing the cost of customized algorithm development.

[0037] This invention deeply couples the communication status awareness mechanism of IGV with the lane change / departure control algorithm to construct a dual-mode adaptive control framework of "networked master control + network-off redundancy". When communication is good, it obtains the global optimal information through vehicle-road cooperation to achieve efficient coordination; when communication is interrupted, it independently completes environmental perception, parking space detection and path planning by relying on the on-board perception system and local cached data, ensuring uninterrupted operation and greatly improving the robustness and availability of the system.

[0038] In the network-disconnected redundancy mode, independent parking space detection is achieved based on active scanning of lidar without relying on external information. Combined with short-time high-precision pose estimation of the integrated inertial navigation system, accurate lane changing and parking space exit are completed. At the same time, a smooth filtering transition mechanism and a status synchronization breakpoint resume mechanism are designed to ensure that mode switching is shock-free and task continuity is uninterrupted, which has extremely high engineering practical value in actual port deployment. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 This is a flowchart provided for an embodiment of the present invention; Figure 2 This is a schematic diagram of the dynamic safe approach distance for oblique motion provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the initial state of de-queueing provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the oblique translation and turning posture before de-queueing provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of the oblique translation and turning posture after exiting the formation, provided in an embodiment of the present invention. Figure 6 This is a schematic diagram of the oblique movement before insertion provided in an embodiment of the present invention.

[0041] The following are the labeling elements in the figure: 1. IGV vehicle; 2. Obstacle ahead; 3. Obstacle behind; 4. Current lane centerline; 5. Target lane centerline; 6. Target parking space boundary.

[0042] The accompanying drawings illustrate specific embodiments of the invention, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the invention in any way, but rather to illustrate the concept of the invention to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0043] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0044] To make the technical solutions and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0045] Example 1: Control of reciprocating oblique exit from narrow parking spaces Please see Figures 1 to 5 As shown in the figure, this embodiment provides a complete implementation process of the lane change and narrow parking space exit control method for IGV in the narrow parking space exit scenario.

[0046] In this embodiment, IGV vehicle 1 is parked in a yard parking space, and the total length of the vehicle is [missing information]. It is 16 meters long and the width is... The length is 3.2 meters, the wheelbase is 9.6 meters, and the distance from the vehicle's center of gravity to the front axle is... It is 4.8 meters, the distance from the vehicle's center of gravity to the rear axle. The distance is 4.8 meters. Obstacle 2 is 4.8 meters from the front of the vehicle. The distance is 8.0 meters, and the obstacle 3 behind the vehicle is 3 meters away from the rear of the vehicle. The current lane width is 2.0 meters. There is an empty space in the target lane on the right, and the target is exiting in the right direction. The target lane width is 4.0 meters. The distance is 4.0 meters. Communication quality is good, and the vehicle is operating in network-controlled mode.

[0047] First, scene perception and task type identification steps are performed. The integrated navigation system provides the vehicle's current pose information, including the vehicle's x and y coordinates in the geodetic coordinate system, the vehicle's heading angle θ, and the front axle equivalent rotation angle. Equivalent rotation angle of rear axle The lidar detected an obstacle 2 at a distance. The distance is 8.0 meters, and the distance to the obstacle behind is 3. The ultrasonic sensor has a range of 2.0 meters and performs precise distance measurement and verification at close range. Based on the presence of obstacles both in front and behind and insufficient longitudinal space, the system determines the current operation scenario to be a vehicle exiting a confined space.

[0048] The feasibility of de-lineup is then determined. The system bases this on the distance 2 to the obstacle ahead. 3 distances from the obstacle behind A dual-condition admission decision is made. The first condition is sufficient longitudinal space, requiring that the total usable space in front and behind the vehicle is greater than the sum of the vehicle's total length and safety redundancy. ,in The spatial coefficient is set to 1.2 in this embodiment. The static safety redundancy distance is taken as 0.5 meters. The calculation yields... = 10.0 meters, and = 16×1.2 + 0.5 = 19.7 meters. 10.0 meters is less than 19.7 meters, so the condition of sufficient longitudinal space is not met.

[0049] At this point, a pre-adjustment step to the intermediate position is performed. The system generates a yaw moment in place through differential steering of the front and rear axles, causing the vehicle to make slight attitude adjustments within the parking space to redistribute the front and rear space margins. Specifically, the control system sends opposite steering angle commands to the front and rear axles, causing the vehicle to rotate around the vertical axis with almost no longitudinal displacement, thereby adjusting the vehicle to the center position within the parking space. After the pre-adjustment is completed, the distance to obstacle 2 ahead is re-detected. It becomes 8.5 meters, with a distance of 3 meters from the obstacle behind. It became 1.5 meters. + The longitudinal space requirement remains at 10.0 meters, which is still insufficient. The system determines that leaving the vehicle is not feasible, triggers the safety protection mechanism, keeps the vehicle stationary, and reports an "insufficient space" status.

[0050] If the longitudinal space condition is met in another scenario, the system proceeds to determine the second condition—the insufficient forward space trigger condition. This condition requires the forward space to be less than the minimum forward safe distance threshold d required to complete a full lane change in diagonal mode. thd1 ,Right now < . The calibration method is as follows: ,in This represents the maximum permissible deviation angle of the vehicle's heading relative to the lane direction in diagonal driving mode; in this embodiment, it is set to 12°. For a safety margin, we take 1.2. Substitute the values ​​into the calculation: =(4.0×16) / (16×sin12° + 3.2×cos12°)×1.2. sin12°≈0.208, cos12°≈0.978, the denominator is 16×0.208 + 3.2×0.978 = 3.328 + 3.130 = 6.458, the numerator is 64.0, 64.0 / 6.458≈9.91, then multiply by the safety factor of 1.2 to get... ≈11.9 meters. If currently... If the distance is 8.0 meters, which is less than 11.9 meters, the forward space is insufficient, and the trigger condition is met. If both conditions are met, the system enters the reciprocating oblique de-line mode.

[0051] After entering the reciprocating diagonal de-queue mode, the system calculates the dynamic safe approach distance in real time during each diagonal movement. , The calculation formula is: in The current longitudinal velocity, The system response delay time is set to 0.5 seconds in this embodiment. For the maximum braking deceleration, take 2.5 m / s². The minimum safe clearance after parking is set to 0.2 meters. The target vehicle speed is set at 3.0 km / h, which is approximately 0.833 m / s. Substituting into the formula, the calculation is as follows: = 0.833×0.5 + 0.833² / (2×2.5) + 0.2 = 0.417 + 0.139 + 0.2 = 0.756 meters.

[0052] To further improve security, It is also corrected based on the road surface adhesion coefficient μ. The correction formula is: in To correct for the strength coefficient, a value of 0.25 is used. If the current road surface adhesion coefficient... The value is 0.85 (dry road surface). The correction term is (1-0.85) / (0.85+0.5)×0.25 = 0.15 / 1.35×0.25≈0.028. After correction... =0.756 × 1.028 ≈ 0.777 meters. If the road surface is slippery, If the value is 0.4, then the correction term is (1-0.4) / (0.4+0.5)×0.25 =0.6 / 0.9×0.25≈0.167. After correction... = 0.756 × 1.167 ≈ 0.882 meters.

[0053] The vehicle then proceeds diagonally towards the target direction. It switches to diagonal mode and travels diagonally towards the target direction, i.e., to the right front. The system uses an omnidirectional diagonal kinematic model to control the vehicle. The expression for this model is: , , , in( , () represents the coordinates of the vehicle reference point in the geodetic coordinate system. For the vehicle's heading angle, The sideslip angle is the angle at the vehicle's center of gravity. The front axle equivalent rotation angle, This is the equivalent rotation angle of the rear axle. By deflecting the front and rear axles in the same direction, the sideslip angle of the vehicle's center of gravity approaches zero, resulting in minimal overall translation and yaw motion, which is within an acceptable range.

[0054] Desired front axle steering angle and rear axle desired steering angle The generation method is as follows: in This refers to the target diagonal direction angle, relative to the lane direction. In this embodiment, the target diagonal direction angle... Let the angle be 12°. Substituting the values, we get: tan12°≈0.213. = arctan(9.6×0.213 / (9.6+4.8×0.213)) = (2.045 / 10.622) = (0.1925)≈10.9°, = (4.8 × 0.213 / 9.6) = (1.022 / 9.6) = (0.1065)≈6.1°. The vehicle is moving diagonally to the right at a speed of 3.0 km / h, and the distance to obstacle 2 ahead is detected in real time. The vehicle was brought to a stop when it descended to 0.756 meters. The distance traveled uphill was recorded. It is 0.85 meters.

[0055] Next, switch to the step of moving diagonally backwards towards the target direction. The target direction angle for the backward diagonal movement is the opposite of the forward diagonal direction angle, i.e. The angle is -12°. During the backward tilting process, the system performs adaptive convergence control on the vehicle's heading deviation. Attitude deviation. = - ,in This is the direction angle of the target lane centerline. Currently... The deviation is 11.2°, which is greater than the preset attitude deviation threshold. (In this embodiment, 5° is used). The system generates a corrected yaw moment by adjusting the difference in rotation angle between the front and rear axles. The control law is: in and To maintain the basic turning angle required for the current diagonal direction, To control the gain proportionally, we set it to 0.15. The differential control gain is set to 0.05. The rate of change of heading deviation is currently approximately 0.2° / s. Substitute this into the calculation: = 10.9° + 0.15×11.2° + 0.05×0.2 = 10.9° + 1.68° + 0.01° = 12.59° = 6.1° - 1.68° - 0.01° = 4.41°. The vehicle diagonally moves to the right rear while gradually correcting its course. During the diagonal movement, the distance to the obstacle 3 behind is monitored in real time. _current Braking was initiated when the distance reached 0.756 meters, and the subsequent diagonal distance was recorded. It is 0.60 meters.

[0056] After completing one full cycle of "forward diagonal movement + backward diagonal movement", the system calculates the lateral deviation of the current vehicle relative to the center line 5 of the target lane. and the longitudinal cumulative displacement of the vehicle that has moved out of its original parking space . This is the perpendicular distance from the vehicle reference point to the straight line containing the center line 5 of the target lane. Longitudinal cumulative displacement. The calculation formula is: = , where n is the number of loops completed. and Δ These represent the actual distances traveled during the forward and backward diagonal movements in the i-th iteration, respectively. Let be the average diagonal direction angle of the i-th cycle, taken as its absolute value. In this embodiment, after completing one cycle... =(0.85 + 0.60)×sin12° = 1.45×0.208≈0.302 meters. Minimum displacement required to exit. Calculate using the formula below. in To determine the safe clearance after entering the target lane, take 0.2 meters, and substitute it into the equation. = 3.2 / 2 + 4.0 / 2 + 0.2 = 1.6 + 2.0 + 0.2 = 3.8 meters. Lateral deviation The distance is 1.22 meters, which is greater than the preset threshold. (In this embodiment, the length is 0.2 meters), and the conditions for exiting the formation are not met.

[0057] The system performs single-cycle displacement monitoring. If the actual longitudinal displacement increment Δ in a certain cycle... Less than the preset minimum effective displacement threshold A value of 0.5 times the value of the vehicle indicates that the vehicle has entered a state of stationary oscillation. In this embodiment, Δ... It is 0.302 meters. Taking a value of 0.3 meters, 0.302 meters is greater than 0.15 meters, so it has not fallen into oscillation and continues the cycle. Current cycle number. The maximum allowed number of loops is 1. Dynamically set according to parking space. = =14 times. Return to continue executing the next iteration of the loop.

[0058] After 12 cycles Reaching 3.85 meters, greater than 3.8 meters, The deviation is 0.08 meters, less than 0.2 meters, indicating the current heading deviation. | is 1.2°, which is less than the preset threshold (3° in this embodiment). When all three conditions are met, the system determines that the vehicle has completely left the original parking space and successfully entered the target lane, exits the reciprocating diagonal driving mode, returns vehicle control to the regular path tracking module, resumes the regular driving mode, and completes the exit operation.

[0059] Example 2: Lane Change and Positioning Control Please see Figure 6 As shown, this embodiment provides a complete implementation process for a lane-changing and parking scenario.

[0060] In this embodiment, IGV vehicle 1 is driving normally in the current lane (right lane) at a speed of The speed is 10 km / h, approximately 2.78 m / s. There are available parking spaces in the target lane on the left. The length of the parking space is... The distance is 18 meters. Current position. Given (0, 0, 0°), the target pose P end (60 meters, -3.8 meters, 0°).

[0061] First, the system identifies available parking spaces in the target lane. The vehicle performs a gridded scan along the lane direction within the target lane area, dividing the target lane into multiple 0.1m x 0.1m grid units. The point cloud density of the LiDAR within each grid is calculated. Continuous grid areas with a point cloud density below a preset threshold (less than 3 points per grid in this embodiment) are identified as available parking spaces. After scanning and identification, a continuous available grid area was found at a longitudinal position between 50m and 68m. The length of the available parking space is 18m, which is greater than the sum of the vehicle's total length of 16m and the safe clearance of 0.5m, thus meeting the parking requirements.

[0062] The feasibility of the lane change is then assessed. The system calculates the Collision Risk Index (CRI) of the lane change path, using the formula CRI = , where m is the number of obstacles within the perception range. Let j be the shortest distance from the j-th obstacle to the candidate lane change path. The distance influence factor is set to 1.0 meter. Calculations show that there are no obstacles within 3 meters of the candidate lane change path, resulting in a CRI value of 0.12, which is less than the preset threshold. (In this embodiment, we take 0.5) and determine that there are no obstacles in the lane change path, and trigger the lane change process.

[0063] Next, a smooth lane change path is generated, using the vehicle's current position as the starting point for the lane change. =(0, 0, 0°), with the center of the available parking space (59 meters longitudinally and -3.8 meters laterally) as the endpoint for lane changing. =(59m, -3.8m, 0°), a smooth lane change path is generated using a fifth-order polynomial. The lane change start point is located on the current lane centerline 4, and the lane change end point is located within the target parking space boundary 6 on the target lane centerline 5.

[0064] The generated fifth-order polynomial path is discretized into 118 path points at 0.5-meter intervals. The vehicle tracks these path points in a diagonal pattern. Within each control cycle (50 milliseconds in this embodiment), the lateral deviation from the vehicle's current position to the path is calculated. and heading deviation The desired steering angles of the front and rear axles are calculated using a kinematic model, causing the vehicle to travel diagonally along the path. The target speed is set at 3.0 km / h, and the distance to obstacles ahead is detected in real time, employing a dynamic safe approach distance. Perform a safety braking assessment. The calculation method is the same as in Example 1.

[0065] As the vehicle travels diagonally along the path towards the target parking space, it switches to a backward diagonal movement in the target direction. The target direction angle for the backward diagonal movement is the opposite of the forward diagonal direction angle. The system adjusts the vehicle's heading to align with the center line 5 of the target lane using an attitude deviation control law. During the backward diagonal movement, the distance to the obstacle 3 behind is monitored in real time to achieve... Stop diagonally when the time comes.

[0066] After completing one cycle of "forward diagonal movement + backward diagonal movement", calculate the lateral deviation of the current vehicle relative to the centerline 5 of the target lane. If the lateral deviation is less than... (0.2 meters) If the vehicle has completely entered the target parking space boundary 6, it is determined that the deviation is within 0.2 meters. If this condition is not met and the maximum number of cycles has not been reached, the forward and backward diagonal driving operations are repeated. In this embodiment, after 3 cycles, the lateral deviation drops to 0.08 meters, which is less than 0.2 meters, and the parking is considered complete. The system exits the diagonal driving mode and resumes the normal driving mode.

[0067] Example 3: Autonomous Control in Network Disconnection Redundancy Mode This embodiment illustrates the autonomous network disconnection control capability of the present invention in the event of a communication interruption.

[0068] During an IGV's unloading operation in a confined space, the 5G communication signal suddenly dropped. Dual communication link monitoring showed that the main link's RSSI was below -100dBm, the signal-to-noise ratio (SNR) was below 5dB, and it remained unrecovered for 2.5 seconds. Four consecutive heartbeat packets were lost, indicating a network outage. The system immediately switched to network redundancy mode.

[0069] The vehicle control system automatically loads the latest yard map, lane geometry information, and parking space layout data from its local cache. The perception system switches to a pure vehicle-mounted perception mode, relying solely on LiDAR, ultrasonic sensors, and a combined inertial navigation system for environmental perception. The onboard human-machine interface displays a "Network disconnection mode in operation" message, and an audible and visual alarm issues a network disconnection warning.

[0070] In the offline redundancy mode, target lane vacancy confirmation does not rely on V2X communication, but is entirely achieved through active scanning and detection by LiDAR. The LiDAR performs a 360-degree scan of the target lane at a frequency of 10 Hz to acquire point cloud data. The system then scans the boundaries on both sides of the target lane, clusters the obstacle point clouds within the target area, identifies the start and end boundaries of the vacant space, and calculates its usable length along the lane direction. In this embodiment, the lidar detected that there was available space in the right lane. The distance is 20 meters, which is greater than the sum of the vehicle's total length of 16 meters and the minimum safe clearance of 0.5 meters, so it is determined that it is safe to drive in.

[0071] Dynamic safe approach distance Conservative parameters are used for calculation in offline mode. Increased to 1.0 second (considering system response degradation during network outages). Reduced to 1.8 m / s², Increased to 0.4 meters. The target speed remains 3.0 km / h. = 0.833×1.0 + 0.833² / (2×1.8) + 0.4 = 0.833 + 0.193 + 0.4 = 1.426 meters. Compared to 0.756 meters in the network-connected mode, the safe distance in the offline mode increases by approximately 88%, ensuring sufficient braking safety margin.

[0072] The maximum number of loops is reduced to 60% to 80% of the number in offline mode. This embodiment uses online mode. 14 times, offline mode = (14 × 0.6) = 9 times. Before each cycle begins, the current spatial state is reassessed using a lidar. If the assessment result indicates that a single diagonal movement is sufficient to complete the de-coupling (lateral deviation)... Less than 2× If there is sufficient vertical space, then single diagonal decoupling will be prioritized to reduce the number of cycles and shorten the network outage time.

[0073] The vehicle performs reciprocating diagonal exit control while in a network-disconnected state. The specific procedure is the same as in Example 1, but conservative parameters are used in the network-disconnected mode. After 7 cycles, the vehicle successfully leaves its original parking space and enters the target lane. The continuous operation time in the network-disconnected mode is approximately 75 seconds, which does not exceed the safe operation time limit. (This example is set to 180 seconds).

[0074] After communication was restored, the system detected that the 5G link had returned to normal, and the Q value rose to a stable level and remained above [a certain value]. (4.0 seconds in this embodiment). The vehicle will upload the operation status recorded during the network outage (including current pose, number of completed cycles of 7, cumulative displacement, and final parking position) to the fleet management system through the restored communication link.

[0075] When switching back from the offline redundancy mode to the network-connected main control mode, the system uses a first-order low-pass filter to smoothly transition control commands. The fleet management system reassesses the current operation progress based on the synchronized status information, determines that the vehicle has successfully left the formation, and does not need to continue the departure operation. It then directly issues subsequent driving commands, achieving breakpoint resume.

[0076] Example 4: Adaptive Change of Dynamic Safety Distance and Correction of Road Surface Adhesion Coefficient This embodiment illustrates dynamic safe approach distance. Adaptive variation characteristics under different vehicle speeds and road surface conditions.

[0077] Set system response delay time Maximum braking deceleration is 0.5 seconds. The safe parking clearance is 2.5 m / s². The distance is 0.2 meters. Under dry road conditions, the road adhesion coefficient μ is 0.85, and the corrected strength coefficient λ is 0.25.

[0078] When the vehicle speed is 1.0 km / h (approximately 0.278 m / s), = 0.278×0.5 + 0.278² / (2×2.5) +0.2 = 0.139 + 0.015 + 0.2 = 0.354 meters. Corrected for road surface adhesion coefficient: = 0.354×(1 + (1-0.85) / (0.85+0.5)×0.25) = 0.354×(1 + 0.15 / 1.35×0.25) = 0.354×1.028≈0.364 meters.

[0079] When the vehicle speed is 3.0 km / h (approximately 0.833 m / s), = 0.833×0.5 + 0.833² / (2×2.5) +0.2 = 0.417 + 0.139 + 0.2 = 0.756 meters. (Corrected) ≈0.777 meters.

[0080] When the vehicle speed is 5.0 km / h (approximately 1.389 m / s), = 1.389×0.5 + 1.389² / (2×2.5) +0.2 = 0.695 + 0.386 + 0.2 = 1.281 meters. (Corrected) ≈1.317 meters.

[0081] Under wet and slippery road conditions, the road adhesion coefficient μ is 0.4, and other parameters remain unchanged. At a vehicle speed of 3.0 km / h, the correction term is (1-0.4) / (0.4+0.5)×0.25 = 0.6 / 0.9×0.25≈0.167. = 0.756 × 1.167 ≈ 0.882 meters, which is about 13.5% more than the 0.777 meters on a dry road surface. At a speed of 8.0 km / h, =2.299×1.167≈2.683 meters, an increase of about 16.7%.

[0082] The above data shows that when the vehicle speed increases from 3.0 km / h to 5.0 km / h, The braking distance increased from 0.756 meters to 1.281 meters, an increase of approximately 70%, ensuring braking safety at high speeds; while at low speeds of 1.0 km / h... With a distance of only 0.354 meters, this avoids vehicles stopping prematurely when there is still space due to excessively large safety distances. When the road surface adhesion coefficient decreases from 0.85 to 0.4, The increase of approximately 13% to 17% reflects the safety compensation for the increased braking distance on low-adhesion road surfaces. This adaptive characteristic is the core advantage of the dynamic safety distance calculation in this invention.

[0083] Example 5: Extreme operating condition application of the intermediate position pre-adjustment mechanism This embodiment illustrates the working process of the intermediate position pre-adjustment mechanism under extremely confined space conditions.

[0084] Overall length of vehicle The distance is 16 meters, and the obstacle in front is 2 meters away. The distance is 3.0 meters, and the distance to the obstacle behind is 3. It is 2.0 meters long, with a total usable space. + = 5.0 meters. Spatial coefficient α is taken as 1.2, static safety redundancy distance d safe0 Taking 0.5 meters as an example, the requirement for sufficient longitudinal space is as follows. + >16 × 1.2 + 0.5 = 19.7 meters. 5.0 meters is much smaller than 19.7 meters, so directly entering the reciprocating diagonal de-lineation mode is not feasible.

[0085] At this point, the pre-adjustment step for the intermediate position is triggered. The system generates a yaw moment in place through differential steering between the front and rear axles: the front axle yaws 10° to the left and the rear axle yaws 10° to the right, allowing the vehicle to rotate around the vertical axis with almost no longitudinal displacement. During the pre-adjustment process, the system monitors the distances to obstacle 2 ahead and obstacle 3 behind in real time at a frequency of 5 Hz, ensuring that at any moment... and All safety boundaries shall be no less than 0.3 meters.

[0086] After approximately 3 seconds of pre-adjustment, the vehicle's posture rotated by about 8°, and its position within the parking space shifted from slightly rearward to near the center. The distance d_f of obstacle 2 ahead was re-detected and became 3.5 meters, while the distance of obstacle 3 behind was... It became 3.5 meters. + = 7.0 meters. 7.0 meters is still less than 19.7 meters, so the condition of sufficient longitudinal space is still not met.

[0087] If the system determines that leaving the formation is not feasible, it will trigger a safety protection mechanism. The vehicle will remain stationary and report a status of "insufficient space, unable to leave the formation safely". At the same time, it will alert on-site personnel through audible and visual alarms and wait for manual intervention or remote dispatch instructions.

[0088] This embodiment illustrates the rigor of the dual-condition access logic of the present invention—it will neither mistakenly deny dequeueing when there is sufficient space, nor risk entering and causing a collision when there is severely insufficient space. The intermediate position pre-adjustment mechanism provides a final opportunity for improvement in extreme operating conditions. When the pre-adjustment still cannot meet the safety conditions, the system actively shuts down instead of blindly trying, reflecting the safety design concept of the present invention.

[0089] Example 6: Bidirectional Symmetrical Adaptive Left De-line Control This embodiment illustrates the bidirectional symmetrical adaptation capability of the present invention, taking the scenario of de-coiling to the left as an example.

[0090] Symmetrical to the scenario of exiting to the right in Example 1, the vehicle is parked in a yard parking space, there is an empty space in the target lane on the left, and the target vehicle is exiting in the left direction. Distance to obstacle 2 ahead. The distance is 8.5 meters, with obstacles 3 meters behind. The width is 2.5 meters, and the total longitudinal space is 11.0 meters, which meets the requirement of sufficient longitudinal space. thd1 The calibration result is the same as in Example 1, approximately 11.9 meters. =8.5 meters is less than 11.9 meters, the forward space is insufficient, the trigger condition is met, and the reciprocating diagonal de-line mode is entered.

[0091] When moving to the left, all direction parameters are completely symmetrical to those when moving to the right. In step S23, the forward diagonal movement in the target direction becomes a forward diagonal movement to the left, and the target diagonal direction angle... Set to -12° (12° to the left relative to the lane direction). Desired front axle steering angle: Desired rear axle steering angle: The direction of the turn is opposite to that when exiting the line to the right.

[0092] In step S24, the direction is switched to leftward and diagonally backward, with a target orientation angle of +12°. In the attitude deviation control law... and The symbol remains unchanged, but the symbol of the control command is adjusted accordingly because the target direction is reversed.

[0093] The loop determination logic, threshold conditions, and maximum loop count settings in steps S25 to S27 are exactly the same as those for decoupling to the right. After approximately 12 loops, the lateral deviation... The longitudinal cumulative displacement decreased to 0.08 meters, less than 0.2 meters. Reaching 3.85 meters, greater than 3.8 meters, heading deviation | | If the angle is 1.5° or less than 3°, the de-lineup is complete.

[0094] This embodiment demonstrates that the present invention supports bidirectional adaptation throughout the entire process, by adjusting the target direction angle θ. target The left or right exit operation can be achieved by reversing the direction of rotation, without modifying the core control algorithm. This symmetrical design allows the invention to flexibly adapt to the lane changing and exit requirements of different port yard layouts and different operating directions.

[0095] Example 7: Verification of Multi-Source Sensor Fusion and Three-Level Security Protection This embodiment illustrates the working process of the multi-source sensor fusion architecture and three-level security protection mechanism of the present invention.

[0096] The IGV is equipped with a differential GPS / INS integrated navigation system, a 16-line LiDAR, ultrasonic sensors, and a V2X vehicle-to-everything (V2X) communication unit. The four sensing channels operate in parallel, enabling redundant detection and cross-verification of pose and obstacle information.

[0097] During normal operation, the integrated navigation system provides centimeter-level positioning information, lidar is responsible for mid-to-long-range obstacle detection, ultrasonic sensors are responsible for close-range fine ranging, and V2X communication acquires yard maps and real-time parking space occupancy information broadcast by roadside units. These four types of information are fused after time synchronization and spatial coordinate unification, and are then mutually verified.

[0098] If the lidar fails at a single point, the system relies on ultrasonic sensors for nearby obstacle information, V2X for parking space information, and integrated navigation for pose information to still safely complete the operation. If V2X communication is interrupted, the system switches to a network-disconnected redundancy mode and operates independently using the onboard perception system.

[0099] A three-tiered security protection mechanism that progresses step by step: The first layer is dynamic safety distance real-time braking. In each control cycle (50 milliseconds), the system compares the real-time detected obstacle distance with the dynamic safety approach distance. When d obs ≤ The braking stop command is triggered immediately. In this embodiment, during a certain forward diagonal movement... When the distance decreases from 1.5 meters to 0.756 meters, the braking response time is approximately 80 milliseconds (including a 40-millisecond perception delay and a 40-millisecond actuator response), and the vehicle comes to a safe stop at a distance of approximately 0.2 meters from the obstacle.

[0100] The second layer monitors the displacement over a single cycle. In the 6th cycle, the actual longitudinal displacement increment... Δ The value is 0.12 meters, which is less than the preset minimum effective displacement threshold. If the distance is 0.5 times (0.3 meters), i.e. 0.15 meters, the system determines that the vehicle has entered a state of stationary oscillation, immediately terminates the reciprocating cycle and triggers the safety protection mechanism, the vehicle brakes to stop and reports "oscillation abnormality".

[0101] The third layer is a protection against the maximum number of loops. If the number of loops reaches... (In this embodiment, the number of attempts is set to 14) If the vehicle still fails to exit the formation, the system triggers the safety protection mechanism, the vehicle brakes and stops, and an abnormal status of "exit timeout / insufficient space" is reported.

[0102] The three-layer protection system progresses step by step, from real-time braking to oscillation detection and then to timeout protection, ensuring that the system can stop safely under any abnormal operating conditions and completely eliminating collision accidents.

[0103] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention. This application is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the following claims.

[0104] It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A method for controlling lane changing and reciprocating oblique exiting from tight parking spaces using an IGV (Incoming Vehicle), characterized in that, Includes the following steps: S1: Scene perception and operation type recognition, obtain IGV pose information, obstacle distance information and target lane status information, and determine whether the current operation scenario is a narrow parking space exit scenario or a lane change into parking scenario. S2: When the scenario is determined to be a narrow parking space exit, execute the reciprocating diagonal exit control process, including: S21: Based on the distance to obstacles ahead Distance to rear obstacles The feasibility of exiting the formation is determined. If both the conditions of sufficient vertical space and insufficient forward space are met, the reciprocating oblique exiting mode is entered. S22: During each diagonal movement, a safe approach distance is dynamically set based on the current vehicle speed. When the distance to the obstacle is less than or equal to the stated distance... When the brakes are triggered, the vehicle stops. S23: Move diagonally forward in the target direction, control the vehicle to move forward horizontally, and detect the distance to obstacles ahead in real time. When the target is reached... Stop at this time; S24: Switch to diagonal movement towards the target direction, perform adaptive convergence control on the vehicle's heading deviation, and detect the distance to obstacles behind in real time. When the target is reached... Stop at this time; S25: After completing one "forward diagonal movement + backward diagonal movement" cycle, calculate the lateral deviation of the vehicle relative to the centerline of the target lane. and longitudinal cumulative displacement ; S26: When the lateral deviation The longitudinal cumulative displacement is less than a preset threshold and When the minimum displacement required to exit is reached, the exit from the formation is considered complete; S27: If the de-queueing is not completed and the current loop count has not reached the upper limit, return to step S22 to continue the loop; if the loop count upper limit is reached, trigger the safety protection mechanism. S3: When the scenario is determined to be a lane change into a parking space, the lane change into a parking space control procedure is executed, including: S31: Identify available parking spaces in the target lane; S32: Determine the feasibility of changing lanes in the available parking spaces. If it is feasible, trigger the lane-changing process. S33: Generate a smooth lane change path with the vehicle's current position as the starting point of the lane change and the center of the available parking space as the ending point of the lane change. S34: Proceed diagonally forward along the radial target direction of the changed road, using the same dynamic safe approach distance as in step S22. Perform a safety braking assessment; S35: Switch to diagonal movement towards the target direction and perform adaptive convergence control on the vehicle's heading deviation; S36: Calculate the lateral deviation of the vehicle relative to the center line of the target lane. If the lateral deviation is less than a preset threshold, the vehicle is determined to be in position. If the vehicle is not in position and the number of cycles has not reached the upper limit, repeat steps S34 to S35. S37: Exit the diagonal mode when the infeed condition is met or the maximum number of loops is reached.

2. The lane changing and narrow parking space reciprocating oblique exit control method for IGV according to claim 1, characterized in that: The longitudinal space sufficiency condition mentioned in step S21 is that the total front and rear space is greater than the sum of the total vehicle length and the safety redundancy; the forward space insufficiency trigger condition is that the forward space is less than the minimum forward safety distance threshold d required to complete a full lane change in diagonal mode. thd1 The d thd1 The calibration is based on a combination of lane width, vehicle dimensions, maximum diagonal deviation angle, and safety factor. If the aforementioned sufficient longitudinal space condition is not met, a pre-adjustment of the intermediate position is performed: a yaw moment is generated by differential steering of the front and rear axles, causing the vehicle to make a small-amplitude attitude adjustment within the parking space to redistribute the torque. and Then make a judgment again; If the pre-adjustment conditions for sufficient longitudinal space are still not met, decoupling is deemed infeasible, triggering a safety protection mechanism.

3. The lane changing and narrow parking space reciprocating oblique exit control method for IGV according to claim 1, characterized in that: The dynamic safe approach distance mentioned in step S22 It is calculated based on the current vehicle speed, system response delay time, and maximum braking deceleration, and is adaptively adjusted in real time according to the vehicle speed.

4. The lane changing and narrow parking space reciprocating oblique exit control method for IGV according to claim 1, characterized in that: The oblique driving mode described in steps S23, S24, S34, and S35 is as follows: by controlling the front axle and the rear axle to deflect in the same direction, the vehicle moves forward or backward as a whole, and the side slip angle of the vehicle's center of gravity approaches zero.

5. The lane changing and narrow parking space reciprocating oblique exit control method for IGV according to claim 1, characterized in that: The adaptive convergence control of the vehicle heading deviation described in steps S24 and S35 is as follows: detect the deviation between the vehicle heading and the center line of the target lane; when the deviation exceeds a preset threshold, generate a corrective yaw moment by adjusting the difference between the front and rear axle rotation angles, so that the vehicle heading gradually converges to the target lane direction.

6. The lane changing and narrow parking space reciprocating oblique exit control method for IGV according to claim 1, characterized in that: The longitudinal cumulative displacement described in step S25 It is the sum of the longitudinal components of the forward and backward diagonal distances in each iteration.

7. The lane changing and narrow parking space reciprocating oblique exit control method for IGV according to claim 1, characterized in that: Step S25 also includes single-cycle displacement monitoring: if the actual longitudinal displacement increment of a certain cycle is less than the preset minimum effective displacement threshold, it is determined that the vehicle has fallen into a stationary oscillation state, the reciprocating cycle is immediately terminated and the safety protection mechanism is triggered.

8. The lane changing and narrow parking space reciprocating oblique exit control method for IGV according to claim 1, characterized in that: The method for identifying vacant parking spaces on the target lane in step S31 is as follows: perform a gridded scan within the target lane area, count the obstacle point cloud density in each grid, and identify continuous grid areas with point cloud density lower than a preset threshold as vacant parking spaces.

9. The lane changing and narrow parking space reciprocating oblique exit control method for IGV according to claim 1, characterized in that: The lane change feasibility determination in step S32 includes: calculating the length of the available parking space; when the length of the available parking space is greater than the sum of the total length of the vehicle and the safe clearance, further calculating the collision risk index of the lane change path; when the collision risk index is lower than a preset threshold, determining that there are no obstacles in the lane change path and triggering the lane change process. The smooth lane change path described in step S33 is generated using a polynomial, satisfying the continuous constraints of the starting and ending positions, heading angles, and curvature. The maximum curvature of the path is less than the maximum achievable curvature of the vehicle steering system.

10. The lane changing and narrow parking space reciprocating oblique exit control method for IGV according to claim 1, characterized in that: It also includes the communication status awareness and operation mode selection step S0: The IGV is equipped with dual communication terminals and multiple network interfaces to monitor the quality of the communication link in real time. When it is determined that the network is out of service, the vehicle automatically switches to the network outage redundancy mode. When in the network-off redundancy mode, the vehicle loads locally cached map and lane geometry information, and the perception system switches to pure vehicle perception mode. The dynamic safe approach distance Using conservative parameters, the upper limit of the number of cycles is reduced in offline mode; When the communication link is detected to have returned to normal, a smoothing filter is used to transition the control commands.

Images