Vehicle automatic parking method, system, vehicle and electronic device

Abstract

The application relates to the technical field of intelligent vehicles, and discloses a vehicle automatic parking method, a vehicle automatic parking system, a vehicle and an electronic device. The method comprises the following steps: acquiring self-vehicle real-time sensing data and other-vehicle tracking data; performing vehicle parking situation assessment based on the self-vehicle real-time sensing data and the other-vehicle tracking data, and determining a real-time parking situation assessment result; performing parking game decision processing based on the real-time parking situation assessment result, and determining a self-vehicle parking decision result; and controlling the self-vehicle to park based on the self-vehicle parking decision result. Through the method, the competitive scene of vehicle parking can be intelligently evaluated and decided, so that the initiative and efficiency of the parking process are improved under the premise of ensuring safety.

Description

Technical Field

[0001] This application relates to the field of intelligent vehicle technology, and in particular to a method, system, vehicle, and electronic device for automatic parking of a vehicle. Background Technology

[0002] With the development of autonomous driving technology, automated parking has become a key scenario for solving the "last mile" driving problem. Traditional automated parking technology is mostly based on single-vehicle intelligence, and its core assumption is that the parking lot environment is static or cooperative, that is, vehicles will not compete fiercely for the same scarce resource (such as a prime parking space) during the parking process.

[0003] However, with the increasing prevalence of vehicles equipped with automatic parking functions and the scarcity of urban parking spaces, simultaneous operation of multiple vehicles will become the norm, and parking lots will evolve into dynamic, competitive multi-agent environments. Therefore, there is an urgent need for a method that enables vehicles to park efficiently and safely in highly competitive scenarios. Summary of the Invention

[0004] The embodiments of this application are intended to at least partially address one of the technical problems in the related art. To this end, embodiments of this application propose an automatic parking method, system, vehicle, and electronic device.

[0005] Embodiments of this application provide an automatic parking method for a vehicle, the method comprising: acquiring real-time perception data of the vehicle itself and tracking data of other vehicles; performing a parking situation assessment based on the real-time perception data of the vehicle itself and the tracking data of other vehicles, and determining a real-time parking situation assessment result; performing parking game decision processing based on the real-time parking situation assessment result, and determining a parking decision result for the vehicle itself; and controlling the vehicle to park based on the parking decision result for the vehicle itself.

[0006] In some embodiments, a vehicle parking situation assessment is performed based on real-time perception data of the vehicle itself and tracking data of other vehicles to determine the real-time parking situation assessment result, including: performing a vehicle parking situation assessment based on real-time perception data of the vehicle itself and tracking data of other vehicles to obtain the parking spatiotemporal advantage degree of the vehicle itself reaching the target parking space, the parking behavior intention advantage degree of the vehicle itself, and the parking behavior risk prediction value; and determining the real-time parking situation assessment result based on the parking spatiotemporal advantage degree, the parking behavior intention advantage degree of the vehicle itself, and the parking behavior risk prediction value.

[0007] In some embodiments, the real-time perception data of the vehicle includes the vehicle's position data, vehicle speed data, and vehicle heading angle data; the tracking data of other vehicles includes the position data and speed data of other vehicles; the vehicle parking situation assessment is performed based on the real-time perception data and the tracking data of other vehicles to obtain the parking spatiotemporal advantage of the vehicle reaching the target parking space, including: planning a feasible path for the vehicle to reach the target parking space based on the vehicle's position data, vehicle speed data, and vehicle heading angle data, and planning feasible paths for other vehicles to reach the target parking space based on the position data and speed data of other vehicles; calculating the estimated arrival time of the vehicle based on the vehicle's speed, vehicle kinematic constraints, and feasible paths; calculating the estimated arrival time of other vehicles based on the speed, kinematic constraints, and feasible paths of other vehicles; and comparing the time difference between the arrival time of the vehicle and the arrival time of other vehicles to obtain the parking spatiotemporal advantage.

[0008] In some embodiments, a vehicle parking situation assessment is performed based on real-time perception data of the vehicle itself and tracking data of other vehicles to obtain the dominance of the vehicle's parking behavior intention. This includes: extracting a feature vector of another vehicle from the tracking data of other vehicles, wherein the feature vector of another vehicle includes the distance of the other vehicle relative to the target parking space, the speed of the other vehicle, and the difference in heading angle of the other vehicle; performing intention recognition based on the feature vector of the other vehicle to predict the intention of the other vehicle, wherein the intention prediction result of the other vehicle represents the probability that the other vehicle will drive to the target parking space; determining the clear parking intention of the vehicle itself based on the real-time perception data of the vehicle itself; and comparing the clear parking intention of the vehicle itself with the intention prediction result of the other vehicle to determine the dominance of the vehicle's parking behavior intention.

[0009] In some embodiments, the real-time perception data of the vehicle includes the duration of the turn signal activation of the vehicle toward the target parking space and the proportion of the vehicle entering the feasible path of the vehicle; determining the vehicle's explicit parking intention based on the real-time perception data of the vehicle includes: fusing the turn signal activation duration and the proportion of the vehicle's driving based on a preset first weighting coefficient to obtain a quantified explicit parking intention of the vehicle.

[0010] In some embodiments, a vehicle parking situation assessment is performed based on real-time perception data of the vehicle and tracking data of other vehicles to obtain a parking behavior risk estimate, including: predicting the future trajectory of the vehicle and the estimated future trajectory of other vehicles under candidate parking strategies based on real-time perception data of the vehicle and tracking data of other vehicles; performing spatiotemporal conflict detection on the future trajectory of the vehicle and the estimated future trajectory of other vehicles, assessing the collision risk between the trajectories to determine the parking behavior risk estimate.

[0011] In some embodiments, spatiotemporal conflict detection is performed on the future trajectory of the vehicle and the estimated future trajectory of other vehicles to assess the collision risk between the trajectories and determine the parking behavior risk estimate, including: determining the minimum expected collision time and collision probability between the future trajectory of the vehicle and the estimated future trajectory of other vehicles; and fusing the minimum expected collision time and collision probability based on a preset second weighting coefficient to determine the parking behavior risk estimate.

[0012] In some embodiments, parking game decision processing is performed based on real-time parking situation assessment results to determine the parking decision result of the vehicle, including: constructing a dynamic payoff matrix based on real-time parking situation assessment results; and performing hybrid strategy equilibrium calculation based on the dynamic payoff matrix to determine the parking decision result of the vehicle.

[0013] In some embodiments, the autonomous vehicle parking decision result includes a game decision probability value; based on the autonomous vehicle parking decision result, controlling the autonomous vehicle to park includes: mapping the game decision probability value to determine a parking behavior intensity coefficient, and determining a parking behavior based on the parking behavior intensity coefficient; and controlling the autonomous vehicle to park based on the parking behavior.

[0014] In some embodiments, mapping the game decision probability value to determine the parking behavior intensity coefficient, and determining the parking behavior based on the parking behavior intensity coefficient, includes: if the game decision probability value is in a first range, mapping the game decision probability value to obtain a first parking behavior intensity coefficient, and determining a strong suggestion to yield behavior as the parking behavior based on the first parking behavior intensity coefficient; if the game decision probability value is in a second range, mapping the game decision probability value to obtain a second parking behavior intensity coefficient, and determining a tendency to yield behavior as the parking behavior based on the second parking behavior intensity coefficient; if the game decision probability value is in a third range, mapping the game decision probability value to obtain a third parking behavior intensity coefficient, and determining a neutral behavior as the parking behavior based on the third parking behavior intensity coefficient; if the game decision probability value is in a fourth range, mapping the game decision probability value to obtain a fourth parking behavior intensity coefficient, and determining a tendency to proceed behavior as the parking behavior based on the fourth parking behavior intensity coefficient.

[0015] In some embodiments, the method further includes: during the process of controlling the vehicle to park based on the vehicle parking decision result, acquiring vehicle obstacle perception data, and predicting the vehicle's predicted trajectory and the other vehicle's predicted trajectory for a future preset period based on the vehicle's real-time perception data, other vehicle tracking data, and the vehicle's obstacle perception data; performing a real-time driving risk assessment on the vehicle's predicted trajectory and the other vehicle's predicted trajectory, and determining the safety level boundaries and intervention levels of the vehicle's and other vehicle's trajectories.

[0016] In some embodiments, the safety level boundaries include adjustable safety boundaries, warning boundaries, and emergency boundaries; the intervention levels include Level 1 intervention, Level 2 intervention, and Level 3 intervention; real-time driving risk assessment is performed on the predicted trajectories of the self-vehicle and other vehicles to determine the safety level boundaries and intervention levels of the self-vehicle and other vehicle trajectories, including: if the predicted trajectories of the self-vehicle and other vehicles reach the adjustable safety boundary, Level 1 intervention is used to optimize and adjust the parking behavior of the self-vehicle; if the predicted trajectories of the self-vehicle and other vehicles reach the warning boundary, Level 2 intervention is used to replan the parking behavior of the self-vehicle; if the predicted trajectories of the self-vehicle and other vehicles reach the emergency boundary, Level 3 intervention is used to trigger the emergency braking safety protocol of the self-vehicle.

[0017] An embodiment of this application provides an automatic parking system for a vehicle, comprising: a first determining module, used to assess the vehicle's parking situation based on real-time perception data of the vehicle itself and tracking data of other vehicles, and to determine the real-time parking situation assessment result; a second determining module, used to perform parking game decision processing based on the real-time parking situation assessment result, and to determine the vehicle's parking decision result; and a control module, used to control the vehicle to park based on the vehicle's parking decision result.

[0018] An embodiment of this application provides a vehicle including a memory and a processor. The memory stores a computer program, characterized in that the processor executes the computer program to implement the steps of the method of any of the above embodiments.

[0019] Embodiments of this application provide an electronic device, which includes: a memory, and one or more processors communicatively connected to the memory; the memory stores instructions executable by the one or more processors, which are executed by the one or more processors to cause the one or more processors to implement the steps of the method of any of the above embodiments.

[0020] Embodiments of this application provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method of any of the above embodiments.

[0021] Embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the steps of the method according to any of the above embodiments.

[0022] The automatic parking method provided in this application integrates real-time perception data of the vehicle itself and tracking data of other vehicles. The system can accurately assess the real-time parking situation, determine the real-time parking situation assessment result, and perform parking game decision processing based on the determined real-time parking situation assessment result. Finally, it determines the vehicle parking decision result that can guide and control the vehicle to park.

[0023] As can be seen, the solution provided in this application, when faced with the typical dynamic scenario of multiple vehicles competing for the same parking space, can intelligently evaluate and make decisions on the competitive scenario through the real-time perception data of the vehicle itself and the tracking data of other vehicles, and obtain the parking decision result of the vehicle itself that conforms to the current parking scenario. Then, based on the parking decision result of the vehicle itself, the vehicle is controlled to park, thereby significantly improving the initiative and efficiency of the parking process and optimizing the overall parking experience while ensuring safety. Attached Figure Description

[0024] Figure 1 A flowchart illustrating an automatic parking method for a vehicle provided in this application embodiment; Figure 2 A schematic diagram of the overall system architecture provided for this application; Figure 3 A schematic diagram of an automatic parking system for vehicles provided in this application embodiment; Figure 4 A block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0025] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0026] With the development of autonomous driving technology, automated parking has become a key scenario for solving the "last mile" driving problem. Traditional automated parking technology is mostly based on single-vehicle intelligence, and its core assumption is that the parking lot environment is static or cooperative, meaning that vehicles will not fiercely compete for the same scarce resource (such as a prime parking space) during the parking process. However, with the increasing prevalence of vehicles equipped with automated parking functions and the scarcity of urban parking spaces, multiple vehicles operating simultaneously will become the norm, and parking lots will evolve into a dynamic, competitive multi-agent environment.

[0027] For example, some technologies lack effective distributed competitive decision-making mechanisms, which can easily lead to vehicles competing for parking spaces when facing the same target parking space. This can cause system deadlocks, such as vehicle standstill, as well as inefficient decision-making, such as ineffective avoidance and repeated planning, and even safety risks.

[0028] Some strategies are essentially "one-off games." For example, the vehicle cannot know the exact intentions of other vehicles; the system can only infer by observing their movements. It lacks the ability to actively compete, negotiate, and make complex decisions, and will interrupt parking if another vehicle occupies the planned route. Its decision-making logic is very simple: as soon as a potential conflict risk is detected, it will proactively withdraw from the competition with "ensuring safety" as the highest principle. While this strategy is safe, it is extremely inefficient, easily leading to vehicles constantly encountering "space-stealing" in busy parking lots and lingering for extended periods, resulting in a poor user experience.

[0029] In summary, some technologies rely on a single avoidance strategy, whose decision-making logic does not take into account multi-vehicle competition. While this approach ensures basic safety, it fails to provide a refined assessment of the competitive landscape, leading to two main drawbacks: First, it easily triggers a "decision deadlock," where multiple vehicles employing conservative strategies stall at the parking space entrance due to mutual deference, impacting traffic efficiency; second, it leads to "frequent replanning," where vehicles easily abandon their target parking spaces and need to repeatedly search and replan their routes, prolonging overall parking time and negatively affecting the user experience.

[0030] Therefore, this application provides an automatic parking method for vehicles. In the typical dynamic scenario of multiple vehicles competing for the same parking space, the method can intelligently evaluate and make decisions on the competitive scenario through real-time perception data of the vehicle and tracking data of other vehicles, and obtain a parking decision result of the vehicle that conforms to the current parking scenario. Then, the method controls the vehicle to park based on the parking decision result, thereby significantly improving the initiative and efficiency of the parking process while ensuring safety.

[0031] Figure 1 This is a flowchart illustrating an automatic parking method for a vehicle, as provided in an embodiment of this application.

[0032] like Figure 1 As shown, the automatic parking method 100 provided in this application includes steps S110-S140.

[0033] Step S110: Obtain real-time perception data of the vehicle and tracking data of other vehicles.

[0034] Real-time perception data for the vehicle itself includes real-time data such as its position, speed, heading angle, and geometric information of the target parking space; tracking data for other vehicles includes data such as the position, speed, and trajectory history of other vehicles. This tracking data is obtained by the vehicle itself through its own sensors (such as vision sensors and lidar) to detect, identify, and track surrounding vehicles.

[0035] Step S120: Based on the real-time perception data of the vehicle itself and the tracking data of other vehicles, perform a vehicle parking situation assessment and determine the real-time parking situation assessment result.

[0036] Based on its own real-time perception data and other vehicle tracking data, the vehicle can evaluate the expected results under different strategies in real time, and thus output a real-time parking situation assessment result that is the optimal strategy based on the current competitive situation.

[0037] Step S130: Based on the real-time parking situation assessment results, perform parking game decision processing to determine the parking decision result of the vehicle.

[0038] For example, a dynamic reward function can be constructed based on the multi-dimensional quantitative results output by the real-time parking situation assessment. This reward function can comprehensively consider the behavior and intentions of other vehicles, and then output the optimal strategy under the current dynamic competitive situation, thereby determining the parking decision result of the vehicle.

[0039] Step S140: Based on the parking decision result of the autonomous vehicle, control the autonomous vehicle to park.

[0040] The result of an autonomous vehicle parking decision is, for example, an instruction to control the vehicle to park according to a certain strategy.

[0041] The automatic parking method provided in this application integrates real-time perception data of the vehicle itself and tracking data of other vehicles. The system can accurately assess the real-time parking situation, determine the real-time parking situation assessment result, and perform parking game decision processing based on the determined real-time parking situation assessment result. Finally, it determines the vehicle parking decision result that can guide and control the vehicle to park.

[0042] As can be seen, the solution provided in this application, when faced with the typical dynamic scenario of multiple vehicles competing for the same parking space, can intelligently evaluate and make decisions on the competitive scenario through the real-time perception data of the vehicle itself and the tracking data of other vehicles, and obtain the parking decision result of the vehicle itself that conforms to the current parking scenario. Then, based on the parking decision result of the vehicle itself, the vehicle is controlled to park, thereby significantly improving the initiative and efficiency of the parking process and optimizing the overall parking experience while ensuring safety.

[0043] In some embodiments of this application, vehicle parking situation assessment is performed based on real-time perception data of the vehicle itself and tracking data of other vehicles to determine the real-time parking situation assessment result. Specifically, the automatic parking system equipped on the vehicle is the core brain for the vehicle's parking decision control and can form an integrated perception and decision control system with the vehicle itself. The system can acquire real-time perception data of the vehicle itself and tracking data of other vehicles from the perception layer, and perform vehicle parking situation assessment based on this data to obtain the vehicle's parking spatiotemporal advantage degree, parking behavior intention advantage degree, and parking behavior risk prediction value; finally, based on the parking spatiotemporal advantage degree, parking behavior intention advantage degree, and parking behavior risk prediction value, the real-time parking situation assessment result is determined.

[0044] Parking spatiotemporal advantage, for example, can be used to quantify the priority of a vehicle relative to other vehicles in reaching a target parking space from the perspective of spatiotemporal consistency. Parking spatiotemporal advantage is a dynamically changing quantity that not only reflects the instantaneous relative position of the vehicles but also integrates the current motion state of both vehicles and the feasible path to the parking space, thereby enabling a more accurate prediction of future developments.

[0045] Parking behavior intention dominance can be quantified, for example, from the perspective of behavioral intention, by measuring the strength and clarity of the intention signals of both parties in a competitive environment. The party with clearer intentions exhibits more predictable behavior and can convey stronger commitment signals in the game, thereby influencing the opponent's decision-making and reducing the risk of double-blind games or conflicts caused by misjudgment of intentions. Parking behavior risk prediction can be used, for example, to quantify the potential collision risk under different competitive strategies, pre-assigning a quantified risk cost to each feasible competitive strategy, thus ensuring safety.

[0046] This application introduces a competitive behavior database and an intent expression mechanism to make vehicle behavior more decisive and predictable. Under the premise of ensuring safety, it can actively occupy parking spaces, thus ensuring the smoothness and reliability of the automatic parking system in real and complex environments.

[0047] In some embodiments of this application, the real-time perception data of the vehicle includes the vehicle's position data, vehicle speed data, and vehicle heading angle data; the tracking data of other vehicles includes the position data and speed data of other vehicles; and the vehicle parking situation assessment based on the real-time perception data of the vehicle and the tracking data of other vehicles to obtain the parking spatiotemporal advantage of the vehicle reaching the target parking space may include: Based on the vehicle's position data, speed data, and heading angle data, a feasible path for the vehicle to reach the target parking space is planned, and a feasible path for other vehicles to reach the target parking space is planned based on their position data and speed data. Based on the vehicle's speed, kinematic constraints, and feasible paths, the estimated arrival time of the vehicle is calculated. Based on the speed, kinematic constraints, and feasible paths of other vehicles, the estimated arrival time of other vehicles is calculated. Based on the time difference between the arrival times of the vehicle and other vehicles, the parking space-time advantage is obtained.

[0048] Specifically, it can target a specific parking space based on the current location data of the vehicle. and his car location data Each method uses a real-time path planning algorithm to generate an optimal or suboptimal feasible path from the current location to the parking space entrance. , This path must satisfy the vehicle's basic kinematic constraints. Then, trajectory optimization is performed on the planned path to generate a time-parameterized trajectory. Subsequently, based on the current speed and acceleration, and considering the vehicle's kinematic performance, such as maximum acceleration and maximum curvature limits, the estimated arrival time required to travel along this trajectory is calculated through integration or sampling prediction. and .

[0049] Parking Space-Time Advantage (ST-Score) is quantified by the following functional relationship:

[0050] Where k is the normalization coefficient, used to adjust the results to a suitable numerical range. Vehicles with smaller ETA values ​​have larger reciprocals and higher dominance scores. When a vehicle's ETA is significantly smaller than that of another vehicle... A significantly positive ST-Score indicates that the vehicle has an absolute advantage in time and space.

[0051] In some embodiments of this application, assessing the vehicle parking situation based on the real-time perception data of the vehicle itself and the tracking data of other vehicles to obtain the dominance of the vehicle's parking behavior intention may include: extracting a feature vector of another vehicle from the tracking data of other vehicles, wherein the feature vector of another vehicle includes the distance of the other vehicle relative to the target parking space, the speed of the other vehicle, and the difference in heading angle of the other vehicle; performing intention recognition based on the feature vector of other vehicles to predict the intention of other vehicles, wherein the intention prediction result of other vehicles represents the probability that the other vehicle will drive to the target parking space; determining the clear parking intention of the vehicle itself based on the real-time perception data of the vehicle itself; and comparing the clear parking intention of the vehicle itself with the intention prediction result of other vehicles to determine the dominance of the vehicle's parking behavior intention.

[0052] You can first predict the intention of the other car and then predict the result. Specifically, the system extracts the state sequence of the other vehicle within the most recent time window, such as the past 3 seconds, from the tracking data to construct its feature vector. Features include, but are not limited to, lateral distance, longitudinal distance, heading angle difference, speed, and lateral speed relative to the target parking space. Then, a lightweight classification model, such as a sequence model based on LSTM (Long Short-Term Memory) or a temporal convolutional network, is used as input, and the model outputs a probability value. This represents the probability that the other car intends to drive towards the target parking space. The higher the value, the better the match between the other vehicle's behavior data and the typical trajectory pattern of driving towards the target parking space.

[0053] The vehicle's real-time perception data determines its clear parking intention. Specifically, for a self-driving vehicle, its parking behavior in this space is definite, but for other vehicles, the behavior is not definite. Therefore, for subsequent game strategy, it is necessary to quantify and simulate the certainty of the self-driving vehicle's behavior relative to other vehicles. In some embodiments, the self-driving vehicle's real-time perception data includes the duration of the self-driving vehicle's turn signal towards the target parking space and the proportion of the self-driving vehicle entering its feasible path; determining the self-driving vehicle's definite parking intention based on the real-time perception data further includes: fusing the turn signal duration and the proportion of the vehicle's movement based on a preset first weighting coefficient to obtain a quantified definite parking intention.

[0054] This method quantifies the degree of clarity regarding the vehicle's parking-related behaviors by using a weighted summation approach. An example of the calculation formula is shown below:

[0055] in, : The duration (in seconds) during which the turn signal is turned on towards the target parking space; : The percentage of vehicles that have entered the designated path for parking in this space (0% to 100%); w1 and w2 are weighting coefficients used to adjust the weight of different behaviors.

[0056] Final output ,in, The higher the value, the more irreversible actions the vehicle has performed specifically for this parking maneuver, and the clearer the other vehicle's intention to take over the parking space.

[0057] The degree of dominance of the driver's parking intention is determined by comparing the clarity of the driver's intention with that of other vehicles.

[0058] in, The score range is [-1, 1]. A positive score indicates that the car had a greater advantage than the other car during the parking process; a negative score indicates that the other car had a greater intention and advantage.

[0059] In the embodiments provided in this application, by fusing real-time perception data of the vehicle and tracking data of other vehicles, the system can plan feasible paths for the vehicle and other vehicles to reach the target parking space and calculate the estimated arrival time, thereby obtaining the parking spatiotemporal advantage by comparing the time differences. Simultaneously, by extracting feature vectors of other vehicles to predict their intentions and quantifying the vehicle's explicit parking intention, the system obtains the parking behavior intention advantage of the vehicle by comparison. The solution provided in this application can achieve accurate quantitative evaluation of parking spatiotemporal advantage and the parking behavior intention advantage of the vehicle, providing comprehensive and objective data support for subsequent game theory decisions, and improving the accuracy and reliability of parking decisions in complex environments where multiple vehicles compete for the same parking space.

[0060] In some embodiments of this application, a vehicle parking situation assessment is performed based on real-time perception data of the vehicle and tracking data of other vehicles to obtain a parking behavior risk estimate, including: predicting the future trajectory of the vehicle and the estimated future trajectory of other vehicles under candidate parking strategies based on real-time perception data of the vehicle and tracking data of other vehicles; performing spatiotemporal conflict detection on the future trajectory of the vehicle and the estimated future trajectory of other vehicles, assessing the collision risk between the trajectories to determine the parking behavior risk estimate.

[0061] In some embodiments, spatiotemporal conflict detection is performed on the future trajectory of the vehicle and the estimated future trajectory of other vehicles to assess the collision risk between the trajectories and determine the parking behavior risk estimate, including: determining the minimum expected collision time and collision probability between the future trajectory of the vehicle and the estimated future trajectory of other vehicles; and fusing the minimum expected collision time and collision probability based on a preset second weighting coefficient to determine the parking behavior risk estimate.

[0062] Specifically, risk scenarios can be simulated first. For example, for each candidate parking strategy, such as acceleration to gain an advantage, constant speed approach, or path blocking, the system will use a vehicle dynamics model to predict the future trajectory of the vehicle and the estimated future trajectory of the other vehicle within a short time domain, such as 3-5 seconds, based on the current state of the vehicle and the other vehicle. For each set of future trajectories of the vehicle and the estimated future trajectories of the other vehicle, key safety indicators are calculated between them and the predicted trajectory of the other vehicle. Key indicators include, for example, the minimum estimated collision time. (Minimum Time to Collision) and Collision Probability (PIC), where the minimum predicted collision time is... This represents the minimum estimated arrival time difference at the intersection of the two vehicle trajectories. The smaller the value, the higher the urgency of the collision. The collision probability (PIC) can take into account prediction uncertainties such as sensor noise and control errors, and calculate the probability that the trajectories of the two vehicles will intersect in space and time using a probabilistic model such as a Gaussian distribution.

[0063] Then, the above key safety indicators are integrated into a comprehensive parking behavior risk prediction value, BR-Score:

[0064] Where α, β, and γ are weighting parameters, calibrated using historical accident data or simulations. This function ensures that when... When the risk score is too low or the PIC is too high, the risk score will increase dramatically.

[0065] In addition, the risk score for the proactive retreat strategy can be set as a baseline (e.g., defined as 0). The BR-Score for other strategies is a relative risk value compared to this baseline.

[0066] In some embodiments, the parking behavior risk prediction, parking spatiotemporal advantage, and vehicle parking behavior intention advantage obtained above can be integrated into a real-time parking situation assessment:

[0067] The vector CV describes the current real-time parking situation assessment result.

[0068] In the embodiments provided in this application, by predicting and comparing the future trajectory of the vehicle under the candidate parking strategy with the estimated future trajectory of other vehicles, the system can perform accurate spatiotemporal conflict detection, thereby quantitatively evaluating the minimum expected collision time and collision probability, and finally merging to generate a parking behavior risk prediction value. This can provide key risk quantification indicators for parking game decision-making, and enhance the reliability of the system's safety assessment and decision-making in complex multi-vehicle competition for the same parking space environment.

[0069] In some embodiments of this application, parking game decision processing is performed based on real-time parking situation assessment results to determine the parking decision result of the vehicle, including: constructing a dynamic payoff matrix based on real-time parking situation assessment results; and performing hybrid strategy equilibrium calculation based on the dynamic payoff matrix to determine the parking decision result of the vehicle.

[0070] First, we define the game theory model, including the set of participants: , Indicates a vehicle, This refers to his car. Each participant... The set of optional strategies is ,in Represents offensive strategy. This represents a concession strategy. (Participants) Benefits The combination of strategies chosen by both parties The decision is made, and the benefit is a dynamic function of the real-time situation assessment CV.

[0071] Based on the real-time parking situation assessment results, the following dynamic revenue matrix is ​​constructed:

[0072] in,

[0073]

[0074] This formula quantifies the benefits of successfully seizing market share, and is positively correlated with the competitive advantage of autonomous vehicles.

[0075]

[0076] This item represents the risk cost of choosing an offensive strategy.

[0077]

[0078]

[0079] This formula quantifies the losses caused by the conflict between the two parties. The higher the spatial and temporal advantage of the other vehicle, the greater the likelihood of conflict and the greater the losses.

[0080]

[0081] in, The constant represents the fixed-time cost-benefit loss incurred due to the concession.

[0082] in, It is a constant, and The , indicates that a stalemate is the least efficient situation.

[0083] His car's revenue function The structure is similar, but the calculation is performed using the real-time situation assessment vector of the other vehicle itself.

[0084] In some embodiments of this application, the autonomous vehicle parking decision result includes a game decision probability value; controlling the autonomous vehicle to park based on the autonomous vehicle parking decision result includes: mapping the game decision probability value to determine a parking behavior intensity coefficient, and determining a parking behavior based on the parking behavior intensity coefficient; and controlling the autonomous vehicle to park based on the parking behavior.

[0085] Before implementing this embodiment, kinematic feasibility verification, dynamic feasibility verification, and environmental collision verification can be performed to ensure that the path curvature does not exceed the vehicle's maximum steering capacity, verify that the acceleration and deceleration are within the tire adhesion limits, and coordinate with the real-time safety verification module provided in this application to perform real-time collision detection.

[0086] Specifically, in each decision cycle, the dynamic payoff matrix can be calculated based on the current real-time situation assessment vector (CV), and then the mixed strategy equilibrium of this matrix can be solved. The solution result is a specific probability value, namely the game decision probability value p∈[0,1], which represents the optimal probability of choosing attack strategy A under the current game situation. The game decision probability value p is output as a continuous decision instruction.

[0087] A standardized library of competitive behaviors can be established first, serving as the basic unit for constructing complex competitive behaviors. This library could include, for example, path planning actions, speed planning actions, and behavioral signal actions. Path planning actions could include standard parking paths along predetermined reference lines, paths moderately offset towards the centerline of the parking space entrance to secure a more advantageous geometric position, stronger path offsets that significantly reduce the space for other vehicles to cut in, and offsets away from the parking space to clearly express the intention to yield. Speed ​​planning actions could include comfortable speeds calculated based on path curvature, moderately increasing tracking speed within a safe range, more obvious acceleration, decreasing tracking speed to express caution, and significantly slowing down or stopping to yield. Behavioral signal actions could include stable intention-expressing behaviors such as keeping turn signals constantly on, and enhanced warning behaviors such as briefly increasing light signals (e.g., hazard lights).

[0088] This application also proposes a decision-behavior intensity mapper that can map the game decision probability value p to behavior intensity coefficients. Mapping function:

[0089] In some embodiments, if the game decision probability value is in the first range, the game decision probability value is mapped to obtain a first parking behavior intensity coefficient, and a strong recommendation to yield behavior is determined as the parking behavior based on the first parking behavior intensity coefficient.

[0090] For example, when the game decision probability value is in the range of 0 ≤ p ≤ 0.3, the intensity coefficient of the first parking behavior is obtained by mapping. Choose the strong yielding combination: clearly express the intention to yield by veering away from the parking space, significantly slowing down or stopping to yield, and keeping the turn signal constantly on as a stable expression of intention.

[0091] In some embodiments, if the game decision probability value is in the second range, the game decision probability value is mapped to obtain a second parking behavior intensity coefficient, and the tendency to yield is determined as the parking behavior based on the second parking behavior intensity coefficient.

[0092] For example, when the game decision probability value is in the range of 0.2 ≤ p ≤ 0.5, the intensity coefficient of the second parking behavior is obtained by mapping. When choosing a transitional combination: linear interpolation between yielding and neutral actions.

[0093] In some embodiments, if the game decision probability value is within a third range, the game decision probability value is mapped to obtain a third parking behavior intensity coefficient, and the neutral behavior is determined as the parking behavior based on the third parking behavior intensity coefficient.

[0094] For example, when the game decision probability value is in the range of 0.4 ≤ p ≤ 0.7, the intensity coefficient of the third parking behavior is obtained by mapping. When choosing a neutral combination: a standard parking path that follows a predetermined reference line, a comfortable speed calculated based on the path curvature, and a stable expression of intent by keeping the turn signals constantly on.

[0095] In some embodiments, if the game decision probability value is in the fourth range, the game decision probability value is mapped to obtain the fourth parking behavior intensity coefficient, and the tendency to pass is determined as the parking behavior based on the fourth parking behavior intensity coefficient.

[0096] For example, when the game decision probability value is in the range of 0.6 ≤ p ≤ 1.0, the intensity coefficient of the fourth parking behavior is obtained by mapping. When necessary, choose an offensive combination: a path that is moderately offset towards the center line of the parking space entrance to occupy a more advantageous geometric position; or a stronger path offset that significantly compresses the space for other vehicles to cut in, and moderately increases the tracking speed within a safe range; or a more obvious acceleration behavior and a short-term enhanced light signal (such as hazard lights) to strengthen the prompt behavior.

[0097] In the embodiments provided in this application, continuous game decision probability values ​​can provide precise and quantifiable inputs for subsequent schemes, effectively controlling the aggressiveness of the final parking action. The embodiments provided in this application parameterize the selected parking behavior and ensure coordinated and consistent actions across different control dimensions through timing synchronization, intensity matching, and smooth transition.

[0098] This application also constructs a competitive decision-making model based on multi-dimensional real-time situational assessment. This model places the vehicle and other vehicles within a dynamic game framework, with the core being the construction of a dynamic payoff function. This function not only considers traditional kinematic parameters such as the relative position and speed of the vehicle and parking space, but also incorporates the predicted probability of the other vehicle's behavioral intentions, the historical continuity of the vehicle's behavior (e.g., previously executed pre-parking actions), and the expected payoffs under different decisions. By calculating the expected outcomes under both preemptive and yielding strategies in real time, the model outputs an optimal decision based on the current competitive situation, enabling the vehicle to flexibly choose whether to persist, cautiously probe, or proactively yield, thereby fundamentally improving the intelligence level and scenario adaptability of the decision-making process.

[0099] As described above, this application designs a multimodal, hierarchical competitive behavior execution strategy to replace simple execution, interruption, and termination operations. This strategy predefines a series of standardized behavioral actions, such as path fine-tuning, speed adjustment, and light signal usage, and establishes precise mapping rules between these actions and the upper-level game decision results, forming a rich library of competitive behaviors. Based on real-time decision-making, the vehicle can dynamically invoke and combine different levels of strategic actions, such as accelerating to block, defensive waiting, or proactive yielding, thereby indirectly and explicitly conveying competitive intent through its own dynamic behavior. This refined behavior control effectively improves the predictability of the vehicle's intent and the flexibility of its strategies during interactions, avoiding misjudgments and inefficiencies caused by a single behavioral pattern.

[0100] This application enables vehicles to make the most advantageous choices based on specific competitive situations through real-time situational assessment and game-theoretic decision-making, significantly reducing unnecessary concessions and the resulting frequent replanning, thereby significantly improving parking efficiency at both the vehicle and system levels.

[0101] Figure 2 A schematic diagram of the overall system architecture provided for this application.

[0102] like Figure 2 As shown, the overall system architecture provided in this application can be integrated into an intelligent parking system within a vehicle, linking perception, decision-making, planning, verification, and control execution into a highly efficient closed loop. The overall system framework includes a perception layer, Module 1 (a competitive decision-making module based on real-time situation assessment), Module 2 (a competitive behavior library and hierarchical execution module), Module 3 (a real-time safety verification module based on model predictive control), and a vehicle control actuator. The perception layer is responsible for collecting information on the vehicle's status, other vehicles' status, and the target parking space. Module 1 includes a real-time situation assessment and game theory decision-making unit, outputting the decision probability P to Module 2. Module 2 has a competitive behavior library and a hierarchical execution module, generating preliminary behavioral instructions and transmitting them to Module 3. Module 3 performs parallel high-frequency rolling verification, trajectory prediction and risk assessment, and hierarchical intervention, outputting behavioral instructions for verification and ultimately generating safety control instructions. Finally, the safety control instructions are sent to the vehicle control actuator to complete vehicle control.

[0103] In some embodiments of this application, the method further includes: during the process of controlling the vehicle to park based on the vehicle parking decision result, acquiring vehicle obstacle perception data, and predicting the vehicle's predicted trajectory and the other vehicle's predicted trajectory for a future preset period based on the vehicle's real-time perception data, other vehicle tracking data, and the vehicle's obstacle perception data; performing a real-time driving risk assessment on the vehicle's predicted trajectory and the other vehicle's predicted trajectory, and determining the safety level boundaries and intervention levels of the vehicle's and other vehicle's trajectories.

[0104] This embodiment can be executed based on the safety real-time verification module of Model Predictive Control (MPC). Details of the safety real-time verification module will be described in detail below.

[0105] The model predictive control (MPC)-based real-time security verification module provided in this application serves as the system's underlying security protection layer, operating independently of the upper-layer decision-making logic. Its core function is to perform forward security verification and real-time optimization on the behavioral trajectories generated by module two, based on the model predictive control method, ensuring that all competing behaviors are executed within strictly defined security boundaries, thereby achieving inherent system security.

[0106] This module operates in parallel and asynchronously with Modules 1 and 2, possessing independent sensor data interfaces and computing units to ensure the real-time nature and independence of obstacle perception and risk assessment. It employs an execution frequency higher than that of the decision-making and control layers (typically 50-100Hz) to re-evaluate and optimize the current planned trajectory within each control cycle.

[0107] like Figure 2 As shown, in the solution provided in this application, a bidirectional data exchange channel is established between the model predictive control-based real-time safety verification module and module two. The real-time safety verification module can forward the trajectory risk level and optimization suggestions, and backward receive behavior adjustment confirmation information. This module adopts a three-level priority design to ensure that safety verification instructions have the highest priority, covering upper-level decision instructions. Simultaneously, the module continuously monitors the real-time system health, tracking key indicators such as sensor data quality, model prediction accuracy, and controller output stability. It is also equipped with a safety event black box to fully record environmental data, decision logic, and intervention effects before and after safety intervention events, providing a reliable basis for post-event analysis and system optimization.

[0108] A unified dynamics model encompassing both the vehicle and other vehicles can be established to predict the predicted trajectories of both vehicles within a short, predetermined timeframe (e.g., 3-5 seconds) based on the current state. Simultaneously, key safety indicators such as time difference, collision probability, and safety margin are calculated. Time difference represents the time difference corresponding to the minimum distance between the predicted trajectories of the vehicle and other vehicles; collision probability, calculated using the Monte Carlo method or Gaussian distribution, considers perception and prediction uncertainties; and the safety margin represents the minimum Euclidean distance maintained between the trajectories.

[0109] This application designs a safety boundary and optimization adjustment scheme, setting adjustable safety boundaries, warning boundaries, and emergency boundaries. Adjustable safety boundaries can define dynamic safety thresholds, adaptively adjusting based on environmental factors such as vehicle speed and road adhesion coefficient; warning boundaries can be adjusted by initiating soft constraints for optimization; and emergency boundaries can be adjusted by triggering hard constraints for immediate intervention. Furthermore, this application provides various correction methods, such as trajectory optimization correction, lateral correction, and longitudinal correction. Trajectory optimization correction can minimize the trajectory while maintaining the behavioral intent when the predicted risk exceeds the warning boundary; lateral correction can fine-tune the path to increase the lateral safety distance; and longitudinal correction can adjust the speed curve to increase the time distance.

[0110] By optimizing the aforementioned multi-objectives, Pareto optimality can be achieved among safety, comfort, and behavior retention, where Pareto optimality means that efficiency has been maximized.

[0111] In some embodiments, an adjustable safety boundary is reached between the predicted trajectory of the autonomous vehicle and the predicted trajectory of other vehicles, and a first-level intervention is used to optimize and adjust the parking behavior of the autonomous vehicle.

[0112] For example, if the predicted trajectory of the vehicle and the predicted trajectory of another vehicle reach the adjustable safety boundary, a Level 1 intervention (optimization adjustment) can be used for adjustment. At this time, the risk level is within the warning boundary, the output trajectory of Module 2 is smoothed and optimized, and the intervention log is recorded.

[0113] In some embodiments, if the predicted trajectory of the vehicle and the predicted trajectory of another vehicle reach a warning boundary, a secondary intervention is used to replan the parking behavior of the vehicle.

[0114] For example, if the predicted trajectory of the vehicle and the predicted trajectory of another vehicle reach the warning boundary, a secondary intervention (forced downgrade) can be used for adjustment, at which point the risk warning line is reached. Exceeding this line triggers proactive and gradual trajectory optimization.

[0115] In some embodiments, if the predicted trajectory of the vehicle and the predicted trajectory of another vehicle reach an emergency boundary, a three-level intervention is used to trigger the emergency braking safety protocol of the vehicle.

[0116] For example, if the predicted trajectory of the vehicle and the predicted trajectory of another vehicle reach an emergency boundary, a three-level intervention (emergency takeover) can be used to make adjustments. When a collision is detected as unavoidable, the underlying safety protocols such as emergency braking are immediately triggered to completely take over vehicle control.

[0117] This application provides a safety verification layer that operates in parallel with the competitive decision-making layer. Based on model predictive control, this layer can predict the trajectories of the vehicle and other vehicles in the short-term future in real time and quantitatively calculate the probability of potential collision risks. When the predicted risk value exceeds a preset safety threshold, this mechanism sends a priority instruction to the behavior execution layer to correct, weaken, or forcibly switch to a safe mode in real time. This design achieves continuous safety monitoring and assurance during the competition process, constraining the proactive competition strategy within acceptable safety boundaries, thereby fundamentally ensuring the safety of the system.

[0118] As can be seen, this application, through an independent security risk verification mechanism, separates security assurance from the decision-making logic and places it at the underlying level, allowing upper-level decisions to more actively pursue efficiency within a security framework. This makes vehicle behavior more reliable in the user's perception, enabling successful parking space grabbing under safe conditions, avoiding long waits or function exits caused by vehicle-related rules, and greatly improving the user experience.

[0119] Figure 3 This is a schematic diagram of an automatic parking system for vehicles provided in an embodiment of this application.

[0120] like Figure 3 As shown, the vehicle automatic parking system 300 provided in the embodiments of this application includes: The acquisition module 310 is used to acquire real-time perception data of the vehicle itself and tracking data of other vehicles.

[0121] The first determining module 320 is used to assess the vehicle parking situation based on the vehicle's real-time perception data and other vehicle tracking data, and to determine the real-time parking situation assessment result.

[0122] The second determining module 330 is used to perform parking game decision processing based on the real-time parking situation assessment results and determine the parking decision result of the vehicle.

[0123] The control module 340 is used to control the vehicle to park based on the vehicle's parking decision results.

[0124] In some embodiments, the first determining module 320 is further configured to: perform a vehicle parking situation assessment based on the vehicle's real-time perception data and other vehicle tracking data to obtain the vehicle's parking spatiotemporal advantage degree, the vehicle's parking behavior intention advantage degree, and the parking behavior risk prediction value; and determine the real-time parking situation assessment result based on the parking spatiotemporal advantage degree, the vehicle's parking behavior intention advantage degree, and the parking behavior risk prediction value.

[0125] In some embodiments, the real-time perception data of the vehicle includes the vehicle's position data, vehicle speed data, and vehicle heading angle data; the tracking data of other vehicles includes the position data and speed data of other vehicles; the first determining module 320 is further configured to: plan a feasible path for the vehicle to reach the target parking space based on the vehicle's position data, vehicle speed data, and vehicle heading angle data, and plan a feasible path for other vehicles to reach the target parking space based on the position data and speed data of other vehicles; calculate the estimated arrival time of the vehicle based on the vehicle's speed, vehicle kinematic constraints, and feasible path; calculate the estimated arrival time of other vehicles based on the speed, kinematic constraints, and feasible path of other vehicles; and compare the time difference between the arrival time of the vehicle and the arrival time of other vehicles to obtain the parking spatiotemporal advantage.

[0126] In some embodiments, the first determining module 320 is further configured to: extract a feature vector of another vehicle from the tracking data of another vehicle, wherein the feature vector of another vehicle includes the distance of the other vehicle relative to the target parking space, the speed of the other vehicle, and the difference in heading angle of the other vehicle; perform intention recognition based on the feature vector of the other vehicle, and predict the intention prediction result of the other vehicle, wherein the intention prediction result of the other vehicle represents the probability that the other vehicle will drive to the target parking space; determine the clear parking intention of the own vehicle based on the real-time perception data of the own vehicle; and compare the clear parking intention of the own vehicle with the intention prediction result of the other vehicle to determine the dominance of the parking behavior intention of the own vehicle.

[0127] In some embodiments, the real-time perception data of the vehicle includes the duration of the turn signal on towards the target parking space and the proportion of the vehicle entering the feasible path of the vehicle; the first determining module 320 is further configured to: fuse the turn signal on duration and the proportion of the vehicle based on a preset first weighting coefficient to obtain a quantified clear parking intention of the vehicle.

[0128] In some embodiments, the second determining module 330 is further configured to: predict the future trajectory of the vehicle and the estimated future trajectory of the other vehicle under the candidate parking strategy based on the real-time perception data of the vehicle and the tracking data of the other vehicle; perform spatiotemporal conflict detection on the future trajectory of the vehicle and the estimated future trajectory of the other vehicle, and assess the collision risk between the trajectories to determine the parking behavior risk estimate.

[0129] In some embodiments, the second determining module 330 is further configured to: determine the minimum estimated collision time and collision probability between the future trajectory of the vehicle and the estimated future trajectory of the other vehicle; and fuse the minimum estimated collision time and collision probability based on a preset second weighting coefficient to determine the parking behavior risk estimate.

[0130] In some embodiments, the second determining module 330 is further configured to: construct a dynamic benefit matrix based on the real-time parking situation assessment results; and perform hybrid strategy equilibrium calculation based on the dynamic benefit matrix to determine the vehicle parking decision result.

[0131] In some embodiments, the vehicle parking decision result includes a game decision probability value; the control module 340 is further configured to: perform mapping calculation on the game decision probability value to determine the parking behavior intensity coefficient, and determine the parking behavior based on the parking behavior intensity coefficient; and control the vehicle to park based on the parking behavior.

[0132] In some embodiments, the control module 340 is further configured to: if the game decision probability value is in a first range, map the game decision probability value to obtain a first parking behavior intensity coefficient, and determine a strong suggestion to yield behavior as a parking behavior based on the first parking behavior intensity coefficient; if the game decision probability value is in a second range, map the game decision probability value to obtain a second parking behavior intensity coefficient, and determine a tendency to yield behavior as a parking behavior based on the second parking behavior intensity coefficient; if the game decision probability value is in a third range, map the game decision probability value to obtain a third parking behavior intensity coefficient, and determine a neutral behavior as a parking behavior based on the third parking behavior intensity coefficient; if the game decision probability value is in a fourth range, map the game decision probability value to obtain a fourth parking behavior intensity coefficient, and determine a tendency to proceed behavior as a parking behavior based on the fourth parking behavior intensity coefficient.

[0133] In some embodiments, the automatic parking system 300 further includes a third determining module, which is used to: acquire obstacle perception data of the vehicle during the process of controlling the vehicle to park based on the vehicle parking decision result, and predict the predicted trajectory of the vehicle and the predicted trajectory of the other vehicle for a future preset period based on the real-time perception data of the vehicle, the tracking data of other vehicles and the obstacle perception data of the vehicle; perform real-time driving risk assessment on the predicted trajectory of the vehicle and the predicted trajectory of other vehicles, and determine the safety level boundary and intervention level of the trajectory of the vehicle and the other vehicle.

[0134] In some embodiments, the safety level boundaries include adjustable safety boundaries, warning boundaries, and emergency boundaries; the intervention levels include Level 1 intervention, Level 2 intervention, and Level 3 intervention; the third determining module is used to: if the adjustable safety boundary is reached between the predicted trajectory of the self-vehicle and the predicted trajectory of another vehicle, use Level 1 intervention to optimize and adjust the parking behavior of the self-vehicle; if the warning boundary is reached between the predicted trajectory of the self-vehicle and the predicted trajectory of another vehicle, use Level 2 intervention to replan the parking behavior of the self-vehicle; if the emergency boundary is reached between the predicted trajectory of the self-vehicle and the predicted trajectory of another vehicle, use Level 3 intervention to trigger the emergency braking safety protocol of the self-vehicle.

[0135] It is understood that for a detailed description of the automatic parking system 300, please refer to the description of the method applied to the automatic parking system above.

[0136] An embodiment of this application provides a vehicle including a memory and a processor. The memory stores a computer program, characterized in that the processor executes the computer program to implement the steps of the method of any of the above embodiments.

[0137] This application provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the method in any of the above embodiments.

[0138] This application provides a computer program product that includes instructions that, when executed by a processor of a computer device, enable the computer device to perform the steps of the method described in any of the above embodiments.

[0139] Figure 4 A block diagram of an electronic device provided in an embodiment of this application.

[0140] This application provides an electronic device, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the method in any of the above embodiments.

[0141] like Figure 4 As shown, for ease of understanding, embodiments of this application illustrate a specific electronic device 400.

[0142] Electronic device 400 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. Electronic device 400 may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0143] like Figure 4 As shown, device 400 includes a computing unit 401, which can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) 402 or a computer program loaded from storage unit 408 into random access memory (RAM) 403. RAM 403 may also store various programs and data required for the operation of electronic device 400. The computing unit 401, ROM 402, and RAM 403 are interconnected via bus 404. Input / output (I / O) interface 405 is also connected to bus 404.

[0144] Multiple components in electronic device 400 are connected to I / O interface 405. These components include: input unit 406, such as a keyboard or mouse; output unit 407, such as various types of displays or speakers; storage unit 408, such as a disk or optical disk; and communication unit 409, such as a network interface card (NIC), modem, or wireless transceiver. Communication unit 409 allows electronic device 400 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0145] The computing unit 401 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 401 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 401 performs the various methods described above. For example, in some embodiments, any one or more of the methods described above can be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 408. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 400 via ROM 402 and / or communication unit 409. When the computer program is loaded into RAM 403 and executed by the computing unit 401, one or more steps of any one or more of the methods described above can be performed. Alternatively, in other embodiments, the computing unit 401 can be configured to perform any one or more of the methods described above by any other suitable means (e.g., by means of firmware).

[0146] It should be noted that the logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this application, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0147] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0148] In the description of this application, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this application, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0149] In the description of this application, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", "axial", "radial", "circumferential", etc., indicating the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application.

[0150] Furthermore, the terms "first," "second," etc., used in the embodiments of this application are for descriptive purposes only and should not be construed as indicating or implying relative importance, or implicitly specifying the number of technical features indicated in this embodiment. Therefore, features defined with terms such as "first" and "second" in the embodiments of this application can explicitly or implicitly indicate that the embodiment includes at least one of those features. In the description of this application, the word "multiple" means at least two or more, such as two, three, four, etc., unless otherwise explicitly and specifically defined in the embodiments.

[0151] In this application, unless otherwise explicitly specified or limited in the embodiments, the terms "installation," "connection," "joining," and "fixing" appearing in the embodiments should be interpreted broadly. For example, a connection can be a fixed connection, a detachable connection, or an integral part; it can also be a mechanical connection, an electrical connection, etc. Of course, it can also be a direct connection, or an indirect connection through an intermediate medium, or it can be the internal communication between two components, or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific implementation.

[0152] In this application, unless otherwise expressly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "on top of," and "over" the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.

Claims

1. A method for automatic parking of a vehicle, characterized in that, The method includes: Acquire real-time perception data of the vehicle itself and tracking data of other vehicles; Based on the real-time perception data of the vehicle and the tracking data of other vehicles, the vehicle parking situation is assessed, and the real-time parking situation assessment result is determined. Based on the real-time parking situation assessment results, parking game decision processing is performed to determine the parking decision result of the vehicle. Based on the parking decision results, the vehicle is controlled to park.

2. The method according to claim 1, characterized in that, The process of assessing the vehicle's parking situation based on the vehicle's real-time perception data and the tracking data of other vehicles, and determining the real-time parking situation assessment result, includes: Based on the real-time perception data of the vehicle and the tracking data of other vehicles, the vehicle parking situation is evaluated to obtain the parking time and space advantage degree of the vehicle reaching the target parking space, the parking behavior intention advantage degree of the vehicle, and the parking behavior risk prediction value. Based on the parking spatiotemporal dominance, the vehicle parking behavior intention dominance, and the parking behavior risk prediction, the real-time parking situation assessment result is determined.

3. The method according to claim 2, characterized in that, The real-time perception data of the vehicle includes its own position data, speed data, and heading angle data; the tracking data of other vehicles includes the position and speed data of other vehicles; based on the real-time perception data and the tracking data of other vehicles, a parking situation assessment is performed to obtain the parking spatiotemporal advantage of the vehicle reaching the target parking space, including: Based on the vehicle's position data, vehicle speed data, and vehicle heading angle data, a feasible path for the vehicle to reach the target parking space is planned; and based on the other vehicle's position data and other vehicle speed data, a feasible path for the other vehicle to reach the target parking space is planned. Based on the vehicle speed, vehicle kinematic constraints, and the vehicle feasible path, calculate the estimated arrival time of the vehicle. Based on the speed of the other vehicle, its kinematic constraints, and its feasible path, calculate the estimated arrival time of the other vehicle. The parking time-space advantage is obtained by comparing the time difference between the arrival time of the vehicle and the arrival time of the other vehicle.

4. The method according to claim 2, characterized in that, Based on the real-time perception data of the vehicle itself and the tracking data of other vehicles, a vehicle parking situation assessment is performed to obtain the dominance of the vehicle's parking behavior intention, including: The feature vector of the other vehicle is extracted from the tracking data of the other vehicle, wherein the feature vector of the other vehicle includes the distance of the other vehicle relative to the target parking space, the speed of the other vehicle, and the difference in heading angle of the other vehicle; Intent recognition is performed based on the feature vector of the other vehicle, and an intent prediction result of the other vehicle is obtained, wherein the intent prediction result of the other vehicle represents the probability that the other vehicle will drive toward the target parking space; Based on the real-time perception data of the vehicle, the vehicle's clear parking intention is determined; The vehicle's explicit parking intention is compared with the predicted intentions of other vehicles to determine the dominance of the vehicle's parking behavior intention.

5. The method according to claim 4, characterized in that, The real-time perception data of the vehicle includes the duration of the vehicle's turn signal activation towards the target parking space and the proportion of the vehicle entering the feasible path of the vehicle; determining the vehicle's explicit parking intention based on the real-time perception data includes: The turn signal activation duration and the driving ratio are fused based on a preset first weighting coefficient to obtain a quantified vehicle parking intention.

6. The method according to claim 2, characterized in that, Based on the real-time perception data of the vehicle itself and the tracking data of other vehicles, a vehicle parking situation assessment is performed to obtain a parking behavior risk prediction value, including: Based on the real-time perception data of the vehicle and the tracking data of other vehicles, the future trajectory of the vehicle and the estimated future trajectory of other vehicles are predicted under the candidate parking strategies. Spatiotemporal conflict detection is performed on the future trajectory of the self-vehicle and the estimated future trajectory of other vehicles to assess the collision risk between the trajectories and determine the estimated risk value of the parking behavior.

7. The method according to claim 6, characterized in that, The step of detecting spatiotemporal conflicts between the future trajectory of the self-vehicle and the estimated future trajectory of other vehicles, and assessing the collision risk between the trajectories to determine the estimated risk value of the parking behavior, includes: Determine the minimum estimated collision time and collision probability between the future trajectory of the vehicle and the estimated future trajectory of the other vehicle; The minimum expected collision time and the collision probability are fused based on a preset second weighting coefficient to determine the estimated risk value of the parking behavior.

8. The method according to claim 1, characterized in that, The parking game decision-making process based on the real-time parking situation assessment results, to determine the parking decision result of the vehicle, includes: Based on the real-time parking situation assessment results, a dynamic revenue matrix is ​​constructed; Based on the dynamic payoff matrix, a hybrid strategy equilibrium calculation is performed to determine the autonomous vehicle parking decision result.

9. The method according to claim 8, characterized in that, The autonomous vehicle parking decision result includes a game-theoretic decision probability value; the step of controlling the autonomous vehicle to park based on the autonomous vehicle parking decision result includes: The game decision probability value is mapped and calculated to determine the parking behavior intensity coefficient, and the parking behavior is determined based on the parking behavior intensity coefficient; Based on the parking behavior, the vehicle is controlled to park.

10. The method according to claim 9, characterized in that, The step of mapping and calculating the game decision probability value to determine the parking behavior intensity coefficient, and determining the parking behavior based on the parking behavior intensity coefficient, includes: If the game decision probability value is in the first range, the game decision probability value is mapped to obtain the first parking behavior intensity coefficient, and the strong suggestion to give way behavior is determined as the parking behavior based on the first parking behavior intensity coefficient. If the game decision probability value is in the second range, the game decision probability value is mapped to obtain the second parking behavior intensity coefficient, and the tendency to give way is determined as the parking behavior based on the second parking behavior intensity coefficient. If the game decision probability value is within the third range, the game decision probability value is mapped to obtain the third parking behavior intensity coefficient, and the neutral behavior is determined as the parking behavior based on the third parking behavior intensity coefficient. If the game decision probability value is in the fourth range, the game decision probability value is mapped to obtain the fourth parking behavior intensity coefficient, and the tendency to pass is determined as the parking behavior based on the fourth parking behavior intensity coefficient.

11. The method according to claim 1, characterized in that, The method further includes: During the process of controlling the vehicle to park based on the vehicle parking decision result, obstacle perception data of the vehicle is acquired, and based on the real-time perception data of the vehicle, the tracking data of other vehicles and the obstacle perception data of the vehicle, the predicted trajectory of the vehicle and the predicted trajectory of other vehicles in the future preset time period are predicted. Real-time driving risk assessment is performed on the predicted trajectory of the self-vehicle and the predicted trajectory of other vehicles to determine the safety level boundaries and intervention levels of the self-vehicle and other vehicle trajectories.

12. The method according to claim 11, characterized in that, The safety level boundaries include adjustable safety boundaries, warning boundaries, and emergency boundaries; the intervention levels include Level 1 intervention, Level 2 intervention, and Level 3 intervention; the real-time driving risk assessment of the predicted trajectory of the self-vehicle and the predicted trajectory of other vehicles to determine the safety level boundaries and intervention levels of the self-vehicle and the other vehicle trajectories includes: If the predicted trajectory of the self-vehicle and the predicted trajectory of the other vehicle reach the adjustable safety boundary, the first-level intervention is used to optimize and adjust the parking behavior of the self-vehicle. If the predicted trajectory of the self-vehicle and the predicted trajectory of the other vehicle reach the warning boundary, the second-level intervention is used to replan the parking behavior of the self-vehicle; If the predicted trajectory of the self-vehicle and the predicted trajectory of the other vehicle reach the emergency boundary, the three-level intervention is used to trigger the emergency braking safety protocol of the self-vehicle.

13. An automatic parking system for vehicles, characterized in that, The system includes: The acquisition module is used to acquire real-time perception data of the vehicle itself and tracking data of other vehicles; The first determining module is used to perform a vehicle parking situation assessment based on the real-time perception data of the vehicle itself and the tracking data of other vehicles, and to determine the real-time parking situation assessment result. The second determining module is used to perform parking game decision processing based on the real-time parking situation assessment results to determine the parking decision result of the vehicle. The control module is used to control the vehicle to park based on the vehicle parking decision result.

14. A vehicle comprising a memory and a processor, said memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1-12.

15. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1-12.

Images