CN-122009158-A - Collaborative auxiliary parking system based on multidimensional sensing data and dynamic safety boundary

CN122009158ACN 122009158 ACN122009158 ACN 122009158ACN-122009158-A

Abstract

The invention relates to the technical field of automatic parking, in particular to a cooperative auxiliary parking system based on multidimensional sensing data and a dynamic safety boundary, which comprises a cooperative sensing module, a trusted verification module, a virtual parking space map building module, a pre-judging dynamic safety boundary module and a cooperative control module. The intelligent parking system comprises a vehicle vision detection module, a virtual parking space map construction module, a pre-judging type dynamic safety boundary module, a cooperative control module and a cooperative control module, wherein the vehicle vision detection module is used for detecting the vehicle vision of a vehicle, the cooperative detection module is used for acquiring historical sensing data of the adjacent vehicle and real-time sensing data of a road side through V2X when the vehicle vision is blocked, the trusted verification module is used for carrying out digital signature verification, cross verification and timeliness verification on multi-source data and outputting observation data with confidence weight, the virtual parking space map construction module is used for generating a virtual parking space map comprising a rigid boundary and a flexible boundary through inverse coordinate transformation and multi-source fusion, the pre-judging type dynamic safety boundary module is used for locking the safety boundary in advance and generating an obstacle avoidance path based on the virtual map before the vehicle vision is blocked, and the cooperative control module is used for executing parking control and dynamically adjusting the boundary. The invention realizes the safety control of the pre-judging obstacle avoidance and layering boundary.

Inventors

ZHENG YU
YANG HUIGUANG
WANG LI

Assignees

西安智通信忠软件技术有限公司

Dates

Publication Date: 20260512
Application Date: 20260410

Claims (10)

1. The collaborative auxiliary parking system based on the multidimensional sensing data and the dynamic safety boundary is characterized by comprising a collaborative sensing module, a trusted verification module, a virtual parking space map construction module, a pre-judging dynamic safety boundary module and a collaborative control module, wherein the collaborative sensing module is used for acquiring historical sensing data of a neighboring vehicle and real-time sensing data of road side equipment through V2X communication when a visual field of the vehicle is blocked by an external obstacle, the trusted verification module is used for carrying out digital signature verification and cross verification on multi-source data acquired by the collaborative sensing module, evaluating the confidence of each data source and outputting obstacle observation data with confidence weight, the virtual parking space map construction module is used for unifying the obstacle observation data with confidence weight to the vehicle coordinate system through inverse coordinate transformation and carrying out multi-source fusion to generate a virtual parking space map comprising a rigid boundary and a flexible boundary, the pre-judging dynamic safety boundary module is used for locking the safety boundary in advance to generate a parking path of the obstacle in the pre-blind area before the vehicle enters the visual field blind area, and the collaborative control module is used for executing parking control according to the parking path generated by the pre-judging dynamic safety boundary module and dynamically adjusting the safety boundary according to the real-time data of the vehicle sensor in the parking process.
2. The collaborative auxiliary parking system based on multi-dimensional sensing data and a dynamic safety boundary according to claim 1, wherein the collaborative sensing module comprises a triggering condition detection submodule, a V2V communication submodule and a V2V communication submodule, wherein the triggering condition detection submodule is used for judging that a target parking space area is possibly positioned right behind a large-sized vehicle when the front-sized vehicle is recognized as the large-sized vehicle with the height of more than or equal to 1.8 m and the distance between the vehicle and the tail of the large-sized vehicle is within a preset parking space length range, triggering the collaborative sensing mode, the data request frame is used for broadcasting a data request frame to surrounding vehicles by adopting a DSRC or C-V2X communication standard, the data request frame comprises a time period of requesting data, a sensor type and requests GPS coordinates, a course angle, a vehicle speed and a yaw rate information of an attached acquisition moment of an adjacent vehicle, and the V2I communication submodule is used for establishing communication connection with a road side intelligent unit deployed at a road node and acquiring obstacle coordinates, types, confidence and time stamps under a global coordinate system output by road side equipment.
3. The collaborative auxiliary parking system based on multidimensional sensing data and dynamic safety boundaries according to claim 1, wherein the trusted verification module comprises a digital signature verification sub-module for verifying the validity of a received digital signature and a certificate chain of a data packet and rejecting the data packet which does not pass verification, a cross verification sub-module for calculating Euclidean distance between obstacle coordinates output by each data source when the same obstacle is observed by a plurality of data sources, marking the distance as high confidence data and giving high fusion weight if the distance is smaller than or equal to a preset threshold value, and marking the distance as low confidence data and giving low fusion weight if the distance is smaller than or equal to a preset threshold value, and an timeliness verification sub-module for checking the difference value between the time stamp of the data packet and the current time of the own vehicle, dividing the data into real-time data, quasi real-time data and historical data according to the difference value and giving different weights respectively.
4. The collaborative auxiliary parking system based on multi-dimensional sensory data and dynamic safety boundaries of claim 1, wherein the virtual parking space map building module comprises an inverse coordinate transformation sub-module for acquiring obstacle coordinates in a neighboring vehicle coordinate system Acquiring global pose of adjacent vehicle at acquisition time Acquiring current global pose of own vehicle Converting the obstacle coordinates in the adjacent vehicle coordinate system into the coordinates in the own vehicle coordinate system through rotation and translation transformation The specific transformation formula is as follows: 。
5. The collaborative auxiliary parking system based on multi-dimensional sensing data and dynamic safety boundaries according to claim 1, wherein the virtual parking space map construction module further comprises a boundary type dividing submodule, the boundary type dividing submodule is used for dividing a region corresponding to a static obstacle with the confidence weight being more than or equal to 0.8 into a rigid boundary according to the confidence weight output by the trusted verification module, the rigid boundary is an absolute non-invasiveness region formed by expanding a first safety distance outside an obstacle outline, the static obstacle with the confidence weight being less than 0.8 or a potential risk region estimated based on historical data is divided into a flexible boundary, the flexible boundary is an early warning region formed by expanding a second safety distance outside the obstacle outline, the second safety distance is larger than the first safety distance, and a dynamic boundary changing along with time is generated according to a predicted track of the dynamic obstacle.
6. The cooperative parking assist system based on multi-dimensional sensor data and dynamic safety boundaries of claim 1, wherein the predictive dynamic safety boundary module comprises an error propagation and adaptive redundancy calculation sub-module for calculating a total error The total error at least comprises the square root of the sum of squares of the distance measurement error of the adjacent car camera, the GPS positioning error of the adjacent car, the GPS positioning error of the own car, the time synchronization error and the calibration error of the sensor, and the safety redundancy is calculated according to the total error Wherein As a reference safety distance, the distance between the two sensors is set to be the reference safety distance, And taking the safety redundancy amount as a reserved space of a dynamic safety boundary, and incorporating the safety redundancy amount into a constraint condition of parking path planning.
7. The collaborative auxiliary parking system based on multi-dimensional sensing data and a dynamic safety boundary according to claim 6, wherein the pre-judging dynamic safety boundary module further comprises a progressive boundary contraction sub-module, wherein the progressive boundary contraction sub-module is used for comparing the position of an obstacle newly observed by a vehicle with the position of the obstacle in the virtual parking space map when the vehicle gradually enters a vision blind area and starts to capture blind area information by a vehicle sensor, confirming that virtual data are accurate if deviation is smaller than or equal to a preset deviation threshold value, reducing the safety redundancy of a corresponding area from a first value to a second value so as to improve space utilization, and reconstructing a local safety boundary based on the vehicle sensor data if the deviation is larger than the preset deviation threshold value.
8. The multi-dimensional sensory data and dynamic safety boundary based collaborative auxiliary parking system of claim 1 wherein the collaborative control module includes a path planning sub-module for employing improved blending The system comprises a pre-judging type dynamic safety boundary module, a real-time tracking and control sub-module, a man-machine interaction sub-module and a control quantity calculation module, wherein the pre-judging type dynamic safety boundary module is used for generating a path point sequence containing position, course angle, curvature and speed suggestions by taking a parking path generated by the pre-judging type dynamic safety boundary module as a global reference, the real-time tracking and control sub-module is used for tracking the path point sequence by adopting a model predictive control algorithm and feeding the dynamic safety boundary as a time-varying inequality constraint back to the control quantity calculation, and the man-machine interaction sub-module is used for displaying the virtual parking space map, the rigid boundary, the flexible boundary and the planned path through a visual interface and sending prompt information when the virtual parking space map, the rigid boundary, the flexible boundary and the planned path need to enter a flexible boundary area.
9. The cooperative parking assistance system based on multi-dimensional sensing data and dynamic safety boundaries of claim 1, further comprising a scene recognition module for recognizing whether a vehicle enters a preset special cooperative parking scene through GPS and high-precision map matching, wherein the special cooperative parking scene comprises an old cell, an underground garage corner and a large-sized vehicle crack area, and the cooperative sensing module is actively activated when the entering of the special cooperative parking scene is recognized and the view of a vehicle is blocked.
10. The collaborative auxiliary parking system based on the multidimensional sensing data and the dynamic safety boundary according to claim 2, wherein the roadside intelligent unit comprises a high-definition image acquisition subunit, an edge calculation subunit, a communication subunit and a time synchronization subunit, wherein the high-definition image acquisition subunit is used for acquiring real-time video streams of a road environment, the resolution is more than or equal to 1080P, the frame rate is more than or equal to 25fps, the edge calculation subunit is used for carrying out target detection and tracking on the real-time video streams, outputting coordinates, types and confidence of obstacles under a global coordinate system, the communication subunit is used for broadcasting processed data to surrounding vehicles in a 5G, 4G or DSRC communication mode, and the time synchronization subunit is used for realizing time synchronization with a vehicle system through GPS/Beidou time service and ensuring that the time stamp precision is less than or equal to 1 millisecond.

Description

Collaborative auxiliary parking system based on multidimensional sensing data and dynamic safety boundary Technical Field The invention relates to the technical field of automatic parking, in particular to a cooperative auxiliary parking system based on multidimensional sensing data and a dynamic safety boundary. Background The existing automatic parking system generally relies on vehicle-mounted sensors (such as cameras, ultrasonic radars and millimeter wave radars) to collect environmental data in real time, and parking is achieved through the steps of space-time calibration, coordinate mapping, obstacle identification, parking space correction and the like. However, such systems have some drawbacks in lateral parking scenarios in that when the void is directly behind a large vehicle (e.g., van, large SUV), the vehicle sensor is able to sense the void profile, but low obstacles (e.g., piers, ground locks, squatting children) in the void that cling to the ground tend to be blocked by the height below the bumper or shadow, outside the effective detection range of the vehicle sensor, resulting in the system failing to identify these hidden obstacles. Meanwhile, the safety boundary of the existing system contracts in real time only after an obstacle is seen, and the existing system belongs to reactive control, lacks of prejudgement capability and is easy to cause collision or parking failure due to perceived lag. Disclosure of Invention Therefore, the invention aims to provide a cooperative auxiliary parking system based on multidimensional sensing data and a dynamic safety boundary, so as to solve the problem that the safety boundary of the existing system contracts in real time only after an obstacle is seen, and the cooperative auxiliary parking system belongs to reactive control and lacks of prejudging capability. The collaborative auxiliary parking system based on the multidimensional sensing data and the dynamic safety boundary comprises a collaborative sensing module, a trusted verification module, a virtual parking space map construction module, a pre-judging dynamic safety boundary module and a collaborative control module, wherein the collaborative sensing module is used for acquiring historical sensing data of a neighboring vehicle and real-time sensing data of road side equipment through V2X communication when a view of the neighboring vehicle is blocked by an external obstacle, the trusted verification module is used for carrying out digital signature verification and cross verification on multi-source data acquired by the collaborative sensing module, evaluating confidence of each data source and outputting obstacle observation data with confidence weight, the virtual parking space map construction module is used for unifying the obstacle observation data with the confidence weight to the coordinate system through coordinate inverse transformation and carrying out multi-source fusion to generate a virtual parking space map comprising a rigid boundary and a flexible boundary, the pre-judging dynamic safety boundary module is used for locking the safety boundary in advance to generate a path of the obstacle in the pre-avoiding blind area before the neighboring vehicle enters the view blind area, and the collaborative control module is used for executing parking control according to the path generated by the pre-judging dynamic safety boundary module and dynamically adjusting the parking boundary data in real time according to the self-parking sensor in the parking process. The cooperative sensing module comprises a triggering condition detection sub-module, a V2V communication sub-module and a V2I communication sub-module, wherein the triggering condition detection sub-module is used for judging that a target parking space area is possibly positioned right behind a large-sized vehicle when the front-sized vehicle is recognized to be the large-sized vehicle with the height of more than or equal to 1.8 m and the distance between the vehicle and the tail part of the large-sized vehicle is within a preset parking space length range, triggering a cooperative sensing mode, the V2V communication sub-module is used for broadcasting a data request frame to surrounding vehicles by adopting a DSRC or C-V2X communication standard, the data request frame comprises a data request time period and a sensor type, and requests GPS coordinates, heading angles, vehicle speeds and yaw rates information of nearby vehicles at the acquisition time, and the V2I communication sub-module is used for establishing communication connection with a road-side intelligent unit deployed at a road node to acquire obstacle coordinates, types, confidence and time stamps under a global coordinate system output by road-side equipment. The trusted verification module comprises a digital signature verification sub-module, a cross verification sub-module and an timeliness verification sub-module, wherein the digital signature verification