Automatic battery replacement guiding method for autonomous vehicle

Abstract

The application relates to an automatic battery replacement guiding method of an automatic driving vehicle, the guiding method comprising the following steps: calling pre-stored vehicle calibration information; calculating a corresponding target parking position according to the vehicle calibration information and specified battery replacement position information carried in a battery replacement instruction; determining a corresponding preparation parking position; acquiring entrance point cloud data at an entrance of a battery replacement station, and determining a corresponding feature reference set according to the entrance point cloud data; controlling the automatic driving vehicle to drive into the battery replacement station from the preparation parking position, and calculating a corresponding real-time position of the automatic driving vehicle; comparing the real-time position with the target parking position until the real-time position meets the accuracy requirement of the target parking position, and then controlling the automatic driving vehicle to stop driving, so that the positioning accuracy decline problem is avoided, and the vehicle can be accurately parked at the battery replacement position under the condition of full automatic driving. The construction and maintenance cost of the battery replacement station can be greatly saved by completely relying on the sensors and computing devices of the vehicle itself.

Description

Technical Field

[0001] This invention relates to the technical field of automatic battery swapping for vehicles, and in particular to an automatic battery swapping guidance method for autonomous vehicles. Background Technology

[0002] Currently, with the development of the logistics and transportation industry, pure electric transport vehicles are gradually being widely used. To ensure the continuous operating time of vehicles, in addition to conventional charging, battery swapping is a common method of replenishment. This involves the vehicle driving into a designated battery swapping station when it needs recharging, parking at the designated location, and communicating with the station to confirm the swap before replacing the onboard battery pack. Battery swapping offers advantages such as high efficiency and fast turnaround time, and is therefore increasingly used in electric transport vehicles.

[0003] Autonomous driving systems typically rely on onboard perception systems to acquire information about the vehicle's operating environment. This information, combined with localization results, is used to control the vehicle's acceleration, deceleration, steering, and obstacle avoidance maneuvers. Common perception sensors include cameras, LiDAR, millimeter-wave radar, and ultrasonic radar. Among these, LiDAR, due to its strong environmental adaptability and high detection accuracy, is the primary sensor used for localization and environmental perception in autonomous vehicles. On the other hand, vehicle localization usually relies on satellite positioning. However, when an autonomous vehicle enters a battery swapping station, satellite signal quality often deteriorates significantly due to obstruction and multipath effects, resulting in a severe degradation in localization accuracy. Since common battery swapping stations require high precision in vehicle parking location, this localization degradation makes it difficult for vehicles to reliably park at the designated swapping location in autonomous driving mode.

[0004] Existing solutions include installing additional LiDAR at battery swapping stations to detect the location of vehicles waiting to be swapped and transmitting the results to the vehicles to assist with their parking. However, this approach requires deploying additional sensors at the swapping stations, increasing construction costs and presenting adaptability issues. Different vehicle models vary in size and other aspects, making it easy to infer location based solely on appearance, especially when there are many different types and numbers of vehicles, thus compromising the universality and accuracy of battery swapping location determination. Summary of the Invention

[0005] To address the problem that autonomous vehicles struggle to accurately park at designated battery swapping locations within battery swapping stations due to decreased satellite positioning accuracy, this application provides an automatic battery swapping guidance method for autonomous vehicles.

[0006] An automatic battery swapping guidance method for an autonomous vehicle, wherein the autonomous vehicle includes at least a satellite positioning device and a lidar, and the automatic battery swapping guidance method correctly guides the autonomous vehicle to a battery swapping station. The automatic battery swapping guidance method for an autonomous vehicle includes:

[0007] When the autonomous vehicle receives a battery swapping instruction, it calls the pre-stored vehicle calibration information and calculates the corresponding target parking location based on the vehicle calibration information and the specified battery swapping location information carried in the battery swapping instruction.

[0008] Based on the target parking location and the signal detection results of the satellite positioning device, the corresponding ready parking location is determined; based on the lidar, the entrance point cloud data at the entrance of the battery swapping station is acquired, and the corresponding feature reference set is determined based on the entrance point cloud data;

[0009] The autonomous vehicle is controlled to drive from the prepared parking position into the battery swapping station. During the journey, the corresponding real-time point cloud data is continuously acquired based on the LiDAR. The real-time point cloud data is matched with the feature reference set to calculate the real-time position of the autonomous vehicle.

[0010] The real-time location is compared with the target parking location. When the real-time location meets the accuracy requirements of the target parking location, the autonomous vehicle is controlled to stop driving and an arrival signal is sent to the battery swapping station for automatic battery swapping.

[0011] By employing the aforementioned technologies, the vehicle can be guided step-by-step to the designated location of the battery swapping station from the moment it receives the battery swapping command, through the coordinated use of onboard satellite positioning devices and lidar. This ensures that autonomous vehicles can achieve precise battery swapping without the support of additional station-end equipment, thereby improving the automation and efficiency of the battery swapping process.

[0012] Preferably, the step of calculating the corresponding target parking location based on the vehicle calibration information and the specified battery swapping location information carried in the battery swapping instruction includes:

[0013] Parse the battery swapping command to extract the corresponding specified battery swapping location information [x] b y b ,θ] T , where θ is the heading angle of the autonomous vehicle;

[0014] Based on the vehicle calibration information, the installation position relationship of the battery pack relative to the autonomous vehicle is determined, and based on the installation position relationship and the designated battery swapping location information [x] b y b ,θ] T Calculate the target parking location [x] that meets the battery swapping requirements. v y v ,θ] T .

[0015] By employing the aforementioned technical means, and by analyzing the battery swapping command and combining it with vehicle calibration information to calculate the target parking position, it can be ensured that the vehicle can always find a precise parking point that meets the battery swapping requirements under different battery pack installation locations and different vehicle models, thereby improving battery swapping adaptability and parking accuracy.

[0016] Preferably, the step of determining the corresponding ready parking location based on the target parking location and the signal detection result of the satellite positioning device includes:

[0017] Based on the signal detection results of the satellite positioning device and / or the pre-stored historical signal quality information, the corresponding distance parameter l is determined;

[0018] Based on the target parking location [x] v y v ,θ] T Based on the distance parameter l, calculate the corresponding ready parking position [x]. p y p ,θ] T , where x p =x v -lcosθ,y p =y v -lsinθ.

[0019] By employing the aforementioned technical means and combining the detection results of satellite positioning signals or historical signal quality information, a reasonable parking position can be determined, enabling vehicles to complete the initial reference for target positioning in areas with good signal strength. This avoids positioning errors caused by satellite signal attenuation within the station and ensures the reliability of subsequent entry guidance.

[0020] Preferably, the step of determining the corresponding feature reference set based on the entry point cloud data includes:

[0021] The entry point cloud data is mapped into the radar coordinate system of the lidar to determine the corresponding set of position coordinates as [x, y, z]. T ∈C0;

[0022] The Hough operation is used to detect the position coordinate set [x, y, z]. T ∈C0 is used for detection to generate standard shape features at the entrance of the battery swapping station. These standard shape features are then integrated to determine the corresponding feature reference set [x, y, z]. T ∈F0.

[0023] By employing the aforementioned technical means, acquiring the entrance point cloud using lidar, and using Hough operations to detect the shape features at the entrance, a stable set of feature references can be generated at the entrance of the battery swapping station. This eliminates the need for subsequent positioning to rely on satellite signals, thereby ensuring high-precision guidance for vehicles even in complex environments.

[0024] Preferably, the standard shape features include the corner position at the entrance of the battery swapping station and the center position of the cross-section of the cylindrical isolation pile.

[0025] By employing the aforementioned technical means, and using structural features such as wall corners or the center of isolation piles as standard shape characteristics, stable and easily identifiable reference objects can be selected in different battery swapping station environments, thereby improving the robustness and adaptability of feature detection and avoiding feature recognition failure due to environmental changes.

[0026] Preferably, the step of controlling the autonomous vehicle to drive from the prepared parking position into the battery swapping station, and continuously acquiring corresponding real-time point cloud data based on the LiDAR during the journey, and matching the real-time point cloud data with the feature reference set to calculate the real-time position of the autonomous vehicle, includes:

[0027] When the autonomous vehicle receives an access signal from the battery swapping station, it is controlled to drive from the prepared parking position into the station. During the journey, the vehicle continuously acquires real-time point cloud data based on the LiDAR, determines the corresponding real-time shape features based on the real-time point cloud data, integrates the various real-time shape features, and then determines the corresponding monitoring feature set [x]. ′ y ′ , z ′ ] T ∈F t ;

[0028] The monitoring feature set F t The feature benchmark set F0 is matched, and the correspondence between the real-time shape features and the standard shape features is determined by the Hungarian algorithm.

[0029] Based on the aforementioned correspondence, according to the feature benchmark set F0 and the monitoring feature set F t By analyzing the spatial transformation relationship between them, the real-time position of the autonomous vehicle can be calculated.

[0030] By employing the aforementioned technical means, during the process of a vehicle entering a battery swapping station, a set of monitoring features can be acquired in real time using lidar, and the real-time features can be matched with the baseline features using the Hungarian algorithm. This enables dynamic calculation of the vehicle's real-time position, ensuring that the vehicle's trajectory is corrected at any time during its journey, thereby avoiding deviation or error accumulation.

[0031] Preferably, based on the correspondence, the step involves using the feature benchmark set F0 and the monitoring feature set F... t The steps for calculating the real-time position of an autonomous vehicle, based on the spatial transformation relationship between these relationships, include:

[0032] Based on the aforementioned correspondence, the monitoring feature set F is calculated using the iterative nearest point algorithm. t Based on the spatial transformation relationship relative to the reference feature set F0, the corresponding pose transformation parameter set is determined.

[0033] The real-time position of the autonomous vehicle is determined based on the pose transformation parameter set and the prepared parking position.

[0034] By employing the aforementioned technical means, an iterative nearest-point algorithm is introduced based on the matching results to obtain the spatial transformation relationship between the monitoring feature set and the reference feature set, and the pose transformation parameter set is extracted to update the real-time position, thereby improving the accuracy and stability of pose estimation and ensuring the accuracy of vehicle position judgment.

[0035] Preferably, the step of comparing the real-time location with the target parking location until the real-time location meets the accuracy requirement of the target parking location, then controlling the autonomous vehicle to stop driving and sending an arrival signal to the battery swapping station for automatic battery swapping, includes:

[0036] The real-time location is compared with the target parking location to generate a corresponding position heading deviation set [s, d, δ]. T Where s is the displacement component along the forward direction; d is the displacement component along the lateral direction; and δ is the heading angle deviation.

[0037] Determine whether the heading deviation of the position meets the accuracy requirements of the target parking position;

[0038] If so, the autonomous vehicle will stop driving and send an arrival signal to the battery swapping station to perform automatic battery swapping;

[0039] If not, the corresponding front wheel correction is calculated based on the position heading deviation and the real-time vehicle current speed, and the corresponding driving direction is adjusted according to the front wheel correction.

[0040] By employing the aforementioned technical means, and by comparing the real-time position with the target position to generate a position and heading deviation set, a refined determination of the vehicle's status can be achieved. When the error meets the accuracy requirements, the process can be stopped; otherwise, corrections can be made based on the deviation, thereby ensuring that the vehicle can reach the target point with the fewest adjustments and improving battery swapping efficiency.

[0041] Preferably, in the step of calculating the corresponding front wheel correction amount based on the position heading deviation and the real-time acquired vehicle current speed, the formula for calculating the front wheel correction amount steer is:

[0042] steer=w δ ·δ·vw d ·d·v;

[0043] Among them, w δ For heading angle deviation weighting factor; w d is the lateral displacement deviation weighting factor; v is the current driving speed.

[0044] By employing the aforementioned technical means, and by combining the heading angle deviation and lateral displacement deviation with the vehicle speed to calculate the front wheel correction amount, the vehicle can maintain a stable path correction capability under different speed conditions, avoiding control failure caused by excessive speed or excessive deviation, thereby improving the safety and accuracy of vehicle positioning.

[0045] In summary, this application includes at least one of the following beneficial technical effects:

[0046] In this application, the autonomous vehicle no longer relies on additional sensors deployed at the battery swapping station for positioning assistance. Instead, it utilizes an onboard satellite positioning device, LiDAR, and pre-stored calibration information to form a complete autonomous vehicle guidance mechanism. Specifically, upon receiving a battery swapping command, the vehicle first uses satellite positioning to determine a suitable parking location in a region with good accuracy outside the station. Then, combining the vehicle's calibration information and the battery swapping location information, it calculates the target parking location that meets the battery swapping requirements. Subsequently, the onboard LiDAR collects point cloud data at the station entrance to generate a feature reference set as a positioning reference. During the vehicle's entry into the station, the LiDAR continuously collects real-time point cloud data and matches it with the reference set. Combined with appropriate algorithms, the vehicle's real-time position is obtained, and precise positioning is achieved by comparing it with the target parking location. This avoids the positioning accuracy degradation caused by satellite signal obstruction and multipath effects within the station environment, enabling the vehicle to accurately stop at the battery swapping location under fully autonomous driving conditions. Meanwhile, this application can be realized entirely by relying on the vehicle's own sensors and computing devices, without the need for additional modifications to the battery swapping station or the installation of additional lidar or communication facilities, which greatly saves the construction and maintenance costs of the battery swapping station and improves the overall system's economy and scalability. Attached Figure Description

[0047] Figure 1 This is a flowchart of an automatic battery swapping guidance method for an autonomous vehicle according to one embodiment of this application.

[0048] Figure 2This is a schematic diagram of recording the shape features of the entrance to a battery swapping station in an automatic battery swapping guidance method for an autonomous vehicle according to one embodiment of this application.

[0049] Figure 3 This is a schematic diagram of real-time displacement estimation in an automatic battery swapping guidance method for an autonomous vehicle according to an embodiment of this application. Detailed Implementation

[0050] The present application will be further described in detail below with reference to the accompanying drawings.

[0051] In one embodiment, such as Figure 1 As shown, this application discloses an automatic battery swapping guidance method for an autonomous vehicle. The autonomous vehicle includes at least a satellite positioning device and a lidar. The automatic battery swapping guidance method guides the autonomous vehicle correctly to a battery swapping station. An autonomous vehicle refers to an intelligent vehicle with perception, positioning, planning, and control capabilities, capable of autonomously completing operations such as driving, parking, and obstacle avoidance without human intervention. Its hardware platform typically includes a sensor system, an onboard computing platform, and an execution control system, enabling autonomous decision-making based on environmental information and preset instructions. The satellite positioning device is a positioning module installed on the vehicle, used to receive signals from the Global Navigation Satellite System to calculate the vehicle's position and heading angle in the global coordinate system. Its accuracy is affected by occlusion and multipath effects. The lidar is a sensor that acquires spatial point cloud data of the surrounding environment by emitting a laser beam and receiving the reflected echo. It can achieve high-precision three-dimensional perception of the environmental structure and is commonly used for obstacle detection, distance measurement, and scene feature extraction. The automatic battery swapping guidance method for the autonomous vehicle includes:

[0052] S10. After receiving the battery swapping instruction, the autonomous vehicle calls upon pre-stored vehicle calibration information. Based on the vehicle calibration information and the specified battery swapping location information carried in the battery swapping instruction, it calculates the corresponding target parking location. Vehicle calibration information refers to data pre-measured during the vehicle's manufacturing or operation to describe the relative relationships of key components, such as the fixed installation position relationship between the battery pack and the vehicle's coordinate system. This data is used to subsequently calculate the spatial correspondence between the parking location and the battery swapping location. The specified battery swapping location information is the target parameter sent to the vehicle by the battery swapping station or scheduling system. It specifies the precise location and heading requirements for battery replacement and is a crucial basis for calculating the target parking location. The target parking location is a specific parking point calculated based on the vehicle calibration information and the specified battery swapping location information, meeting the conditions for battery swapping operation and ensuring that the vehicle can successfully complete the battery swapping at that location.

[0053] S20. Based on the target parking location and the signal detection results of the satellite positioning device, determine the corresponding ready parking location; the ready parking location is a temporary reference point determined by satellite positioning outside the battery swapping station. The signal quality at this location is relatively stable and can be used as the starting point for the vehicle to enter the battery swapping station and perform precise positioning.

[0054] S30. Based on the LiDAR, acquire the entrance point cloud data at the entrance of the battery swapping station, and determine the corresponding feature reference set based on the entrance point cloud data. The entrance point cloud data is a set of environmental data collected by the vehicle's LiDAR at the entrance of the battery swapping station, including the three-dimensional coordinate information of structures such as corners and bollards in the entrance area, used to generate the feature reference set. The feature reference set is a set of typical geometric features obtained by processing the entrance point cloud data, serving as a reference benchmark for matching real-time point cloud data when the vehicle enters the battery swapping station. The real-time point cloud data is the dynamic environmental information continuously collected by the LiDAR during the vehicle's entry into the battery swapping station, used for matching with the feature reference set and position estimation.

[0055] S40. Control the autonomous vehicle to drive from the prepared parking position into the battery swapping station. During the journey, continuously acquire the corresponding real-time point cloud data based on the LiDAR. Match the real-time point cloud data with the feature reference set to calculate the real-time position of the autonomous vehicle. The real-time position is the current pose information of the vehicle calculated during the journey, including position coordinates and heading angle, which is used to determine whether the vehicle has reached the target parking position.

[0056] S50. Compare the real-time location with the target parking location until the real-time location meets the accuracy requirements of the target parking location. Then, control the autonomous vehicle to stop driving and send an arrival signal to the battery swapping station to initiate automatic battery swapping. The arrival signal is a confirmation message sent by the vehicle to the battery swapping station after determining that it has stopped at the target parking location, which is used to trigger the start of the battery swapping process.

[0057] Furthermore, the step of calculating the corresponding target parking location based on the vehicle calibration information and the specified battery swapping location information carried in the battery swapping instruction includes:

[0058] S101. Parse the battery swapping command to extract the corresponding specified battery swapping location information [x] b y b ,θ] THere, θ represents the heading angle of the autonomous vehicle. The battery swapping command is an operation request signal issued to the autonomous vehicle by the battery swapping station management system or remote dispatch platform. This command is typically transmitted via a wireless communication link and contains key parameters required for the vehicle to perform a battery swapping operation, such as the battery swapping station number, designated battery swapping location information, and time constraints. It serves as the trigger condition for the vehicle to initiate the automatic battery swapping process. Parsing the battery swapping command refers to the onboard computing equipment performing data parsing and protocol decoding on the received command. This involves structuring the parameter information in the original communication signal and extracting the designated battery swapping location information for subsequent calculation of the target parking location. The designated battery swapping location information refers to the parameters required by the actual battery replacement equipment within the battery swapping station for the vehicle's position and attitude. It typically includes the coordinates and heading angle of the target location, ensuring that the vehicle can accurately dock and disconnect the battery pack after parking at that location.

[0059] S102. Based on the vehicle calibration information, determine the installation position relationship of the battery pack relative to the autonomous vehicle, and based on the installation position relationship and the designated battery swapping location information [x] b y b ,θ] T Calculate the target parking location [x] that meets the battery swapping requirements. v y v ,θ] T Vehicle calibration information is a set of parameters reflecting the spatial relationships between key components, obtained during the production, assembly, or maintenance of autonomous vehicles. This includes information such as the relative position offset of the battery pack installation location to the vehicle coordinate system, interface height, and attitude. This information is acquired through 3D measurement, structural modeling, or factory testing and stored in the onboard control system for geometric calculations and position derivation during operation. The battery pack's installation position relative to the autonomous vehicle refers to its geometric fixation within the vehicle structure, including its displacement in the longitudinal, lateral, and vertical directions, as well as its rotation angles around each axis. Once determined, this relationship remains unchanged throughout the vehicle's lifespan, ensuring the accuracy and mechanical compatibility of battery replacement. By combining the aforementioned installation position relationship with the specified battery replacement location information extracted from the battery replacement command, the vehicle calculation system can derive the target parking location that meets the requirements of the battery replacement operation. This ensures that during actual execution, the vehicle's parking point perfectly matches the position of the battery replacement robot or lifting mechanism, achieving efficient and automated battery pack replacement.

[0060] Furthermore, the step of determining the corresponding ready parking location based on the target parking location and the signal detection result of the satellite positioning device includes:

[0061] S201. Based on the signal detection results of the satellite positioning device and / or pre-stored historical signal quality information, determine the corresponding distance parameter l. The signal detection results refer to the real-time measurement and analysis results of the satellite positioning device on the current satellite signal status during vehicle operation. These results typically include parameters such as the number of visible satellites, signal strength, signal-to-noise ratio, and positioning accuracy factor, used to determine whether the current location meets the requirements for high-precision positioning. Historical signal quality information refers to data on satellite signal reception in a specific area pre-recorded and stored during vehicle operation or debugging. This information can be obtained through multiple samplings, reflecting the signal stability and reliability of a certain area under different time and environmental conditions. The distance parameter l is a spatial offset determined by the signal detection results or historical signal quality information. This parameter represents the preset distance between the target parking position and the intended parking position along the vehicle's direction of travel, ensuring that the intended parking position is in an area with better signal strength, thereby providing a stable reference for accurate positioning within the subsequent battery swapping station.

[0062] S202, Based on the target parking location [x] v y v ,θ] T Based on the distance parameter l, calculate the corresponding ready parking position [x]. p y p ,θ] T , where x p =x v -lcosθ,y p =y v -lsinθ.

[0063] In this embodiment, the vehicle-mounted satellite positioning device performs real-time detection of the signal status of the current environment. The detection indicators include: the number of visible satellites, the average signal-to-noise ratio (SNR), and the positioning accuracy factor. The system presets corresponding thresholds. When the detection results meet the following conditions: the number of satellites is greater than or equal to a preset number threshold, the average SNR is greater than or equal to a preset SNR threshold, and the positioning accuracy factor is less than or equal to a preset accuracy threshold, the current location is determined to have acceptable positioning accuracy. Based on real-time detection, the system also calls the historical signal quality database, which stores the signal strength distribution curve Q(d) of the area outside the battery swapping station, where d represents the distance from the entrance of the battery swapping station. By finding the interval in the function Q(d) that satisfies Q(d)≥Qth, the usable signal stability range [dmin, dmax] is obtained. On the reverse driving path of the vehicle to the target parking position, the satellite positioning device first detects the current signal strength and positioning accuracy in real time. When the detection result shows that the current position still has acceptable accuracy, the system calls the pre-established signal quality database to compare the signal distribution of the area outside the battery swapping station. A point in the middle position is selected in the section where both meet the requirements as the interval between the prepared parking position and the target parking position, thereby obtaining the distance parameter l.

[0064] The vehicle needs to reverse its direction of travel by a displacement of length l to obtain a reference point that maintains a longitudinal distance from the target parking position. This ensures that when the vehicle enters the battery swapping station from its intended parking position along the θ direction, its travel path remains aligned with the target parking position, avoiding repeated corrections due to path deviations. It also ensures that the intended parking position is within a region with reliable satellite signals, facilitating further high-precision positioning of the vehicle using lidar features after entering the station.

[0065] Furthermore, the step of determining the corresponding feature reference set based on the entry point cloud data includes:

[0066] S301. Map the entry point cloud data into the radar coordinate system of the lidar to determine the corresponding position coordinate set as [x, y, z]. T ∈C0; The location coordinate set refers to a set of coordinate points formed by mapping the point cloud data acquired by the lidar at the entrance of the battery swapping station. This set uses the lidar's own coordinate system as a reference system and reflects the spatial distribution of the surfaces of various objects within the entrance area. The lidar coordinate system is a local coordinate reference frame established by the lidar itself, usually with the lidar's installation center as the origin, the forward direction as the x-axis, and the horizontal direction as the y-axis, used to describe and process the acquired point cloud data.

[0067] S302. Using the Hough operation detection method, the set of position coordinates [x, y, z] is analyzed. T∈C0 is used for detection to generate standard shape features at the entrance of the battery swapping station. These standard shape features are then integrated to determine the corresponding feature reference set [x, y, z]. T ∈F0. The Hough operation detection method is a commonly used feature detection algorithm that can transform a set of discrete points in space into a parameter space for cumulative voting, thereby extracting geometric features such as lines and arcs. Therefore, in the processing of entrance point cloud data, it can effectively detect the shape features of geometric structures such as wall edges and isolation posts at the entrance. Standard shape features refer to the idealized features that conform to the entrance geometry identified in the set of position coordinates by the Hough operation detection method.

[0068] In this embodiment, before the vehicle starts and enters the battery swapping station at its intended parking position, it first uses an onboard LiDAR to acquire point cloud data of the entrance area. This point cloud data is then mapped onto a radar coordinate system with the radar installation center as the origin, forming a complete set of position coordinates. Subsequently, the system performs geometric feature detection on this set of position coordinates, employing the Hough operation detection method to map discrete points in space to a parameter space for cumulative statistical analysis, in order to identify geometric structural features conforming to straight lines and arcs. This method can effectively detect standard shape features such as the corner edges of the entrance walls and the centers of the isolation posts. Finally, the system integrates these identified standard shape features to form a feature reference set, which serves as the basis for real-time point cloud matching and pose transformation calculations when the vehicle enters the battery swapping station. This derivation process ensures that the complex point cloud data at the entrance can be simplified into a structured set of geometric features, thus providing reliable input for high-precision vehicle navigation and real-time position estimation.

[0069] Furthermore, the standard shape features include the corner position at the entrance of the battery swapping station and the center position of the cross-section of the cylindrical isolation pile.

[0070] In this embodiment, as Figure 2As shown, the lidar detection features are standard shape features, the lidar data points are data points in the point cloud data, and the isolation piles are real-world isolation pile entities. After performing Hough operations on the entrance point cloud data, the system identifies and classifies straight line features and circular features. Straight line features correspond to the edge structure of the corner of the battery swapping station entrance, while circular features correspond to the center position of the isolation pile's cross-section. Specifically, after mapping the point cloud data to the radar coordinate system, the system uses Hough linear transformation to find the straight line with the largest voting value in the parameter space, thereby deriving the standard straight line position of the corner edge. Simultaneously, the Hough circle detection method is used to identify circular features within a preset radius range in the point cloud data and calculate their center coordinates, thereby deriving the standard center position of the isolation pile. By identifying and extracting the corner position and the center position of the cylindrical isolation pile, standard shape features at the entrance are formed and incorporated into the feature reference set. This ensures that real-time feature matching and pose calculation can be performed based on these stable geometric references during subsequent vehicle entry into the battery swapping station, improving overall positioning accuracy and robustness.

[0071] Furthermore, the step of controlling the autonomous vehicle to drive from the prepared parking position into the battery swapping station, and continuously acquiring corresponding real-time point cloud data based on the LiDAR during the journey, and matching the real-time point cloud data with the feature reference set to calculate the real-time position of the autonomous vehicle, includes:

[0072] S401. When the autonomous vehicle receives the access signal from the battery swapping station, it controls the autonomous vehicle to drive into the battery swapping station from the prepared parking position. The access signal refers to the authorization information sent to the vehicle by the battery swapping station when it detects that the vehicle is in a condition where the battery swapping process can be performed. This signal is usually transmitted wirelessly and is used to trigger the vehicle to drive into the battery swapping station from the prepared parking position.

[0073] S402. During the movement, based on the lidar, corresponding real-time point cloud data is continuously acquired, and based on the real-time point cloud data, corresponding real-time shape features are determined. These real-time shape features are then integrated to determine the corresponding set of monitoring features [x]. ′ y ′ , z ′ ] T ∈F t Real-time shape features refer to the geometric features extracted from real-time point cloud data acquired by LiDAR during the vehicle's entry into the battery swapping station. These features provide a dynamic description of reference objects such as corners and bollards at the station entrance in relation to the vehicle's current position. The monitoring feature set is a set of geometric references obtained by integrating the real-time shape features. It corresponds to the feature reference set previously generated at the entrance and is used to match and calculate the vehicle's actual pose.

[0074] S403, the monitoring feature set F t The real-time shape features are matched with the feature reference set F0, and the correspondence between the real-time shape features and the standard shape features is determined by the Hungarian algorithm. The correspondence refers to the pairing relationship between each real-time shape feature in the monitoring feature set and the standard shape feature in the feature reference set, which is determined by the algorithm. This relationship ensures that the features acquired at different times and locations can be correctly matched, thus providing a basis for subsequent spatial transformation calculations.

[0075] S404. Based on the aforementioned correspondence, according to the feature benchmark set F0 and the monitoring feature set F t By analyzing the spatial transformation relationship between them, the real-time position of the autonomous vehicle can be calculated.

[0076] In this embodiment, during the vehicle's entry into the battery swapping station, to accurately calculate the real-time location, it is necessary to establish a one-to-one correspondence between the current monitored feature set and the pre-stored feature reference set. Since the number of features in the two sets may differ and there may be spatial discrepancies, an optimization method is needed to find the optimal matching relationship. Specifically, the cost value is calculated for each real-time feature point in the monitored feature set and each standard feature point in the reference set. This cost value can be defined as the Euclidean distance or other geometric difference measure between the two points in the radar coordinate system. The calculated cost matrix reflects the cost of all possible matching pairs.

[0077] Based on this, the Hungarian algorithm is used to solve the cost matrix. The Hungarian algorithm finds the minimum-cost matching method step by step, adjusting the matching relationship in each iteration to ensure that the final correspondence minimizes the overall cost. This ensures that even with some noise points or local occlusion, the main geometric features remain correctly paired. For example, when the monitoring feature set extracts the straight line at the entrance corner and the center of the isolation post, while the reference set contains the standard corner and center positions, the Hungarian algorithm automatically selects the combination with the minimum distance in the cost matrix, ensuring that corner features correspond to corners and center features correspond to centers, thus obtaining a stable and reliable correspondence. This ensures accurate pairing of monitoring features and reference features in dynamic environments, avoiding feature confusion or incorrect association, and providing precise input conditions for subsequent spatial transformation derivation based on the iterative nearest-point algorithm.

[0078] Furthermore, based on the aforementioned correspondence, according to the feature benchmark set F0 and the monitoring feature set F... t The steps for calculating the real-time position of an autonomous vehicle, based on the spatial transformation relationship between these relationships, include:

[0079] S4041. Based on the aforementioned correspondence, the monitoring feature set F is calculated using the iterative nearest point algorithm. tBased on the spatial transformation relationship relative to the reference feature set F0, the corresponding pose transformation parameter set is determined. The iterative nearest point algorithm is an optimization algorithm that solves the spatial transformation relationship between two feature sets by minimizing the distance between point sets. Its basic idea is to pair each feature point in the monitored feature set with the corresponding point in the feature reference set, and continuously adjust the pose parameters through iterative updates until the overall matching error converges to a preset threshold, thereby obtaining a stable spatial transformation relationship. The spatial transformation relationship refers to the mapping relationship between the monitored feature set and the feature reference set in terms of position and direction. It is usually composed of translation and rotation, and is used to describe the corresponding changes between coordinate systems during the process of the vehicle entering the battery swapping station from the ready parking position. The pose transformation parameter set is a set of parameters obtained by quantifying the spatial transformation relationship. Specifically, it includes the translation components of the vehicle in the longitudinal and lateral directions and the rotation component of the heading angle. This parameter set can accurately represent the actual position and attitude of the vehicle relative to the feature reference.

[0080] S4042. Determine the real-time position of the autonomous vehicle based on the pose transformation parameter group and the prepared parking position.

[0081] In this embodiment, to calculate the real-time location of a vehicle based on a set of monitored features as it enters a battery swapping station, the system employs an iterative nearest-point algorithm to derive the spatial transformation relationship between the set of monitored features and the feature reference set. Specifically, an initial correspondence is first established between the set of monitored features acquired by the vehicle from the lidar and the pre-established feature reference set, determined by a preceding Hungarian algorithm. Subsequently, the two sets of corresponding points are input into the ICP algorithm for iterative optimization.

[0082] In each iteration, the system first calculates the Euclidean distance difference between points in the monitored feature set and corresponding points in the reference set, using these differences as an error function. Then, based on the least squares optimization method, it solves for the rigid body transformation parameters that minimize the overall error. These parameters include translation (offset along the x and y axes) and rotation (correction of the vehicle's heading angle) in the two-dimensional scene. Specifically, the computing device obtains the optimal translation and rotation matrices for the current iteration using singular value decomposition (SVD) or least squares closed-form solutions, and uses these as the new pose estimation results. Next, the monitored feature set is transformed using the updated pose parameters to make it closer to the reference set, and then the system proceeds to the next iteration.

[0083] The above iterative process continues until the average distance error between the monitored feature set and the reference set converges to a preset threshold, or the maximum number of iterations is reached. The final output translation and rotation values ​​are integrated into a pose transformation parameter set, which, combined with the ready-to-park position, calculates the vehicle's real-time position.

[0084] For example, in one scenario, the corner point of the entrance wall and the center point of the guardrail are extracted as feature points. Initially, the feature set and the baseline set show a deviation of 0.5 meters and a heading error of 5 degrees. After several iterations of the ICP algorithm, each iteration corrects the translation and rotation parameters, gradually converging the error to a mean square error of 0.02 meters and a heading deviation of less than 0.1 degrees. This allows the vehicle's real-time position to be derived, ensuring its trajectory is continuously corrected and accurately reaches the target parking location.

[0085] Furthermore, the step of comparing the real-time location with the target parking location until the real-time location meets the accuracy requirements of the target parking location, then controlling the autonomous vehicle to stop and sending an arrival signal to the battery swapping station for automatic battery swapping, includes:

[0086] S501. Compare the real-time location with the target parking location to generate a corresponding position heading deviation set [s, d, δ]. T Where s is the displacement component along the forward direction; d is the displacement component along the lateral direction; δ is the heading angle deviation; the position heading deviation set refers to a set of error parameters obtained by comparing the real-time position with the target parking position. This parameter set typically includes the displacement component along the vehicle's forward direction, the displacement component along the lateral direction, and the vehicle's heading angle deviation, used to comprehensively characterize the actual deviation of the vehicle relative to the target position. The displacement component along the forward direction refers to the magnitude of the vehicle's deviation in the longitudinal direction of travel, used to determine whether the vehicle has reached the longitudinal depth of the target position; the displacement component along the lateral direction refers to the vehicle's deviation in the lateral direction, reflecting whether the vehicle has a parallel offset from the target position; the heading angle deviation is the difference in angle between the vehicle's travel direction and the heading angle required by the target parking position, this parameter is directly related to whether the vehicle is directly facing the battery swapping station; for example... Figure 3 As shown, the recorded detection features are the recorded detection features determined before the target parking position, and the real-time detection features are the real-time entry features mentioned above.

[0087] S502. Determine whether the position heading deviation meets the accuracy requirements of the target parking position; the accuracy requirements are a pre-set allowable range used to limit the maximum permissible error of the above three deviation values. When all parameters of the position heading deviation set are within the accuracy requirements range, it is determined that the vehicle meets the parking conditions of the battery swapping position.

[0088] S503. If yes, control the autonomous vehicle to stop driving and send an arrival signal to the battery swapping station for automatic battery swapping; S504. If no, calculate the corresponding front wheel correction amount based on the stated position heading deviation and the real-time acquired vehicle current speed, and adjust the corresponding driving direction according to the front wheel correction amount. The current driving speed is the linear velocity data detected by the speed sensor or on-board controller when the vehicle is in real-time driving mode. This speed parameter is used to calculate the front wheel correction amount when the vehicle does not meet the accuracy requirements, making the vehicle's driving direction and speed adjustment more reasonable.

[0089] Furthermore, in the step of calculating the corresponding front wheel correction amount based on the position heading deviation and the real-time acquired vehicle current speed, the formula for calculating the front wheel correction amount steer is:

[0090] steer=w δ ·δ·vw d ·d·v;

[0091] Among them, w δ For heading angle deviation weighting factor; w d is the lateral displacement deviation weighting factor; v is the current driving speed.

[0092] In this embodiment, the heading angle deviation reflects the angle difference between the vehicle's travel direction and the target direction. If not corrected in time, the vehicle will gradually deviate from the target path during longitudinal travel. Therefore, it needs to be amplified by a weighting factor to prioritize the vehicle's heading correction. The lateral displacement deviation reflects the degree of lateral deviation of the vehicle relative to the target trajectory. When the lateral error is large, without compensation, the vehicle may still be unable to align with the battery swapping station when approaching the target position. Therefore, the formula introduces a lateral deviation component and assigns it an independent weight for balanced control. The travel speed directly determines the vehicle's dynamic response characteristics. When the vehicle is at high speed, the same angle correction will result in greater yaw or trajectory deviation. Therefore, the formula introduces speed as a proportional factor so that the correction amount can be adaptively adjusted with speed changes, thereby avoiding oversteering at high speeds and improving convergence efficiency at low speeds. By combining the heading angle deviation, lateral displacement deviation, and speed, the formula can dynamically output the front wheel correction amount that matches the vehicle's current state, ensuring that the vehicle gradually eliminates position and attitude deviations during travel, achieving a smooth entry into the target parking position. For example, when the vehicle approaches the battery swapping location, if a 2-degree deviation in heading angle and a 0.1-meter lateral deviation are detected, and the current speed is 1 meter per second, the system will calculate a suitable steer value according to the formula, instruct the front wheels to make a slight correction, so that the vehicle gradually aligns with the target position and eventually comes to a stable stop.

[0093] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. An automatic battery swapping guidance method for an autonomous vehicle, characterized in that, The autonomous vehicle includes at least a satellite positioning device and a lidar, and is correctly guided to the battery swapping station using the aforementioned automatic battery swapping guidance method. The automatic battery swapping guidance method for an autonomous vehicle includes: When the autonomous vehicle receives a battery swapping instruction, it calls the pre-stored vehicle calibration information and calculates the corresponding target parking location based on the vehicle calibration information and the specified battery swapping location information carried in the battery swapping instruction. Based on the target parking location and the signal detection results of the satellite positioning device, the corresponding ready parking location is determined; Based on the lidar, the entrance point cloud data at the entrance of the battery swapping station is acquired, and the corresponding feature reference set is determined based on the entrance point cloud data. The autonomous vehicle is controlled to drive from the prepared parking position into the battery swapping station. During the journey, the corresponding real-time point cloud data is continuously acquired based on the LiDAR. The real-time point cloud data is matched with the feature reference set to calculate the real-time position of the autonomous vehicle. The real-time location is compared with the target parking location. When the real-time location meets the accuracy requirements of the target parking location, the autonomous vehicle is controlled to stop driving and an arrival signal is sent to the battery swapping station for automatic battery swapping.

2. The automatic battery swapping guidance method for an autonomous vehicle according to claim 1, characterized in that, The step of calculating the corresponding target parking location based on the vehicle calibration information and the specified battery swapping location information carried in the battery swapping instruction includes: The battery swapping command is parsed to extract the corresponding designated battery swapping location information. ,in, The heading angle of the autonomous vehicle; Based on the vehicle calibration information, the installation position relationship of the battery pack relative to the autonomous vehicle is determined, and based on the installation position relationship and the designated battery swapping location information... Calculate the target parking location that meets the battery swapping requirements. .

3. The automatic battery swapping guidance method for an autonomous vehicle according to claim 2, characterized in that, The step of determining the corresponding ready parking location based on the target parking location and the signal detection results of the satellite positioning device includes: Based on the signal detection results of the satellite positioning device and / or the pre-stored historical signal quality information, the corresponding distance parameter l is determined; Based on the target parking location Calculate the corresponding ready parking position based on the distance parameter l. ,in, .

4. The automatic battery swapping guidance method for an autonomous vehicle according to claim 1, characterized in that, The step of determining the corresponding feature reference set based on the entry point cloud data includes: The entry point cloud data is mapped into the radar coordinate system of the lidar to determine the corresponding set of position coordinates. The set of position coordinates It refers to a set of coordinate point data formed by mapping the point cloud data obtained by lidar at the entrance of the battery swapping station; The Hough operation detection method is used to detect the set of position coordinates. The detection process involves mapping discrete points in space to a parameter space for cumulative statistical analysis to identify geometric structural features conforming to straight lines and arcs. This generates standard shape features for the entrance of the battery swapping station. By integrating these standard shape features, the corresponding feature benchmark set is determined. The feature reference set It serves as a basic reference for real-time point cloud matching and pose transformation calculation when subsequent vehicles enter the battery swapping station.

5. The automatic battery swapping guidance method for an autonomous vehicle according to claim 4, characterized in that, The standard shape features include the corner position at the entrance of the battery swapping station and the center position of the cross-section of the cylindrical isolation pile.

6. The automatic battery swapping guidance method for an autonomous vehicle according to claim 4, characterized in that, The step of controlling the autonomous vehicle to drive from the prepared parking position into the battery swapping station, and continuously acquiring corresponding real-time point cloud data based on the LiDAR during the journey, and matching the real-time point cloud data with the feature reference set to calculate the real-time position of the autonomous vehicle, includes: When the autonomous vehicle receives the access signal from the battery swapping station, it is controlled to drive the autonomous vehicle from the prepared parking position into the battery swapping station. During the movement, the lidar continuously acquires corresponding real-time point cloud data, determines corresponding real-time shape features based on the real-time point cloud data, integrates the various real-time shape features, and then determines the corresponding set of monitoring features. ; The monitoring feature set and the set of feature benchmarks The matching is performed, and the correspondence between the real-time shape features and the standard shape features is determined by the Hungarian algorithm; Based on the aforementioned correspondence, and according to the aforementioned feature reference set and the monitoring feature set By analyzing the spatial transformation relationship between them, the real-time position of the autonomous vehicle can be calculated.

7. The automatic battery swapping guidance method for an autonomous vehicle according to claim 6, characterized in that, Based on the aforementioned correspondence, and according to the feature benchmark set and the monitoring feature set The steps for calculating the real-time position of an autonomous vehicle, based on the spatial transformation relationship between these relationships, include: Based on the aforementioned correspondence, the monitoring feature set is calculated using the iterative nearest point algorithm. Relative to the benchmark feature set Based on the spatial transformation relationship, the corresponding pose transformation parameter set is determined. The real-time position of the autonomous vehicle is determined based on the pose transformation parameter set and the prepared parking position.

8. The automatic battery swapping guidance method for an autonomous vehicle according to claim 1, characterized in that, The step of comparing the real-time location with the target parking location until the real-time location meets the accuracy requirements of the target parking location, then controlling the autonomous vehicle to stop and sending an arrival signal to the battery swapping station for automatic battery swapping, includes: The real-time location and the target parking location are compared to generate a corresponding position and heading deviation set. , where s is the displacement component along the forward direction; d is the displacement component along the lateral direction; This refers to the deviation in heading angle; Determine whether the heading deviation of the position meets the accuracy requirements of the target parking position; If so, the autonomous vehicle will stop driving and send an arrival signal to the battery swapping station to perform automatic battery swapping; If not, the corresponding front wheel correction is calculated based on the position heading deviation and the real-time vehicle current speed, and the corresponding driving direction is adjusted according to the front wheel correction.

9. The automatic battery swapping guidance method for an autonomous vehicle according to claim 8, characterized in that, In the step of calculating the corresponding front wheel correction amount based on the position heading deviation and the real-time acquired vehicle current speed, the formula for calculating the front wheel correction amount steer is: ; in, This is the heading angle deviation weighting factor; is the lateral displacement deviation weighting factor; v is the current driving speed.

Images