Belief estimation and rolling iteration method and system for hybrid traffic cognitive uncertainty
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-30
- Publication Date
- 2026-08-14
AI Technical Summary
[0004]智能汽车可通过车载环境感知设备和通讯设备准确地获取混合交通中其他交通参与者的运动信息，包括加速度、速度、航向、绝对位置、相对位置等客观物理信息，然而其他交通参与者的风格类型则隐式地蕴含在其驾驶行为之中，智能汽车难以显式地通过感知模块直接获取，需要借助经验和对态势的观察做出估计，并在不断迭代中印证自己的信念，信念的构筑也依赖于智能汽车本体的态势理解和知识储备，传统的方法难以通过动态的方式不断刷新智能汽车认知
[0049]经由上述的技术方案可知，与现有技术相比，本发明公开提供了混合交通认知不确定性的信念估计与滚动更迭方法及系统，通过将难以处理的随机不确定性难题纳入智能汽车态势认知体系设计过程，并实现认知理解的自发构筑与滚动更迭，有助于加强智能汽车对混合交通环境的认知与理解，赋予智能汽车类人认知逻辑的优势；同时也有助于提高智能汽车的智能化等级和拟人化程度，以及智能汽车对环境的适应度和融入度，有助于提高智能汽车态势认知的快速性、准确性和鲁棒性，为智能汽车真正自然地融入混合交通生态奠定了基础，对智能汽车技术产业化落地具有十分重要的意义。
Smart Images
Figure CN118469026B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent vehicle situational awareness and estimation technology, and more specifically to a method and system for belief estimation and rolling iteration in the context of mixed traffic cognitive uncertainty. Background Technology
[0002] Currently, intelligent vehicles are a crucial component of intelligent transportation systems and a significant strategic direction for the global automotive industry, poised to lead a new round of technological revolution and industrial transformation. At present, intelligent vehicles with varying levels of intelligence, human-driven vehicles with varying levels of skill, and participants such as cyclists and pedestrians with diverse styles coexist in the transportation system, forming a new form of mixed transportation. Mixed transportation has become an essential path to realizing intelligent transportation systems.
[0003] However, in mixed traffic, intelligent vehicles, human-driven vehicles, and pedestrians are dynamically coupled and continuously interact, exhibiting strong time-varying, random, and uncertain characteristics. This poses new challenges to the situational awareness of intelligent vehicles. Most existing situational awareness methods are based on uniform, rule-based cognitive logic, which is neither effective in dealing with the random uncertainties in mixed traffic nor can it comprehensively improve the overall performance of intelligent vehicles in mixed traffic environments.
[0004] Intelligent vehicles can accurately acquire motion information of other traffic participants in mixed traffic through onboard environmental perception and communication devices, including objective physical information such as acceleration, speed, heading, absolute position, and relative position. However, the style types of other traffic participants are implicitly contained in their driving behavior, which is difficult for intelligent vehicles to obtain explicitly through perception modules. They need to rely on experience and observation of the situation to make estimations and verify their beliefs through continuous iteration. The construction of these beliefs also depends on the intelligent vehicle's own situational understanding and knowledge reserves. Traditional methods are difficult to continuously refresh the intelligent vehicle's cognition in a dynamic way.
[0005] Therefore, overcoming the cognitive inaccuracies, unwavering beliefs, and difficulty in coping with uncertainties in mixed traffic by intelligent vehicles is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention
[0006] In view of this, the present invention provides a method and system for belief estimation and rolling iteration of hybrid traffic cognitive uncertainty, which incorporates the difficult-to-handle random uncertainty problem into the design process of intelligent vehicle situational awareness system, and realizes the spontaneous construction and rolling iteration of cognitive understanding, which helps to improve the speed, accuracy and robustness of intelligent vehicle situational awareness.
[0007] To achieve the above objectives, the present invention adopts the following technical solution:
[0008] Belief estimation and rolling iteration methods for hybrid traffic cognitive uncertainty include:
[0009] Acquire motion data of participants in mixed traffic;
[0010] Determine a first initial belief about cognitive uncertainty and multiple types of events, and randomly distribute the initial belief to the multiple types of events to obtain a second initial belief for each type of event;
[0011] Based on the motion data and the various events, an objective belief in the mixed traffic situation is determined;
[0012] The state variables are divided into intervals and matched with the multiple types of events to obtain the inverse conditional probability model for each type of event.
[0013] Based on the second initial belief, the objective belief, and the reverse inference conditional probability model, the iterative belief for each type of event is obtained, and the global iterative belief is determined as the cognitive result output based on the iterative belief;
[0014] The above process is repeated, assigning the current iterative belief to the initial belief distributed to each type of event in the next iteration stage, until the final cognitive result meets the preset value.
[0015] Preferably, the process of acquiring motion data is as follows:
[0016] The movement information of traffic participants in mixed traffic is collected and recorded through intelligent vehicle-mounted environmental sensing and communication devices.
[0017] The motion data is obtained based on the motion information.
[0018] Preferably, the motion data includes at least: instantaneous acceleration, instantaneous velocity, and instantaneous jerk.
[0019] The state variables include at least: continuous acceleration interval, continuous velocity interval, and continuous jerk interval.
[0020] Preferably, the process for determining the objective belief is as follows:
[0021] Based on the instantaneous acceleration a t Instantaneous velocity v t and instantaneous speed j t Conduct actual parallel observations and determine objective beliefs P(O) for the aforementioned multiple types of events:
[0022]
[0023] Where, π ω N represents any event from multiple event classes. π Represents event π ωThe quantity, P(π) ω ) represents any event π ω The initial belief, P([a t ,v t ,j t ]|π ω ) represents any event π ω The inverse conditional probability model.
[0024] Preferably, the multiple types of events include at least: aggressive events A, cautious events C, experiential events E, and routine events N.
[0025] Preferably, the state variables are divided into intervals, and the specific process is as follows:
[0026] The continuous acceleration interval is divided into 0 and Divide into points;
[0027] The continuous velocity range is determined according to v init and Divide into points;
[0028] The continuous jerky intervals are divided into 0 and Divide into points;
[0029] Among them, a max Indicates the maximum acceleration boundary, a min Denotes the minimum acceleration boundary, v init v represents the initial velocity of traffic participants observed at the start of the first round of belief estimation cycle. max Represents the maximum velocity boundary, v min Denotes the minimum velocity boundary, j max Represents the maximum jerk boundary, j min This indicates the minimum jerkness boundary.
[0030] Preferably, the specific process of matching with the multiple types of events is as follows:
[0031] Match the aggressive event A to [0, a] max Acceleration sub-interval, [v init ,v max Velocity subintervals and [0,j] max ] Rapidity sub-interval;
[0032] Match the aforementioned cautious event C to [a] min [0] acceleration sub-interval, [v min ,v init ] velocity subinterval and [j min [0] abrupt change sub-interval;
[0033] Match the empirical event E to Acceleration sub-interval, velocity sub-intervals and Rapidity sub-interval;
[0034] Match the regular event N to [a] min ,a max Acceleration across the entire range, [v] min ,v max ] velocity over the entire range and [j min ,j max The entire range of urgency.
[0035] Preferably, the iterative belief for each type of event is obtained, specifically including:
[0036] Iterative belief P(A|[a] for aggressive event A) t ,v t ,j t ])for:
[0037] Iterative belief P(C|[a] for a cautious event C t ,v t ,j t ])for:
[0038] Iterative belief P(E|[a] for empirical event E t ,v t ,j t ])for:
[0039] Iterative belief P(N|[a] for routine events N t ,v t ,j t ])for:
[0040] Wherein, P([a t ,v t ,j t ]|A) represents the inverse conditional probability model of aggressive event A, P(A) represents the second initial belief of aggressive event A, P([a t ,v t ,j t ]|C) represents the inverse conditional probability model of the cautious event C, P(C) represents the second initial belief of the cautious event C, P([a t ,v t ,j t ]|E) represents the inverse conditional probability model of empirical event E, P(E) represents the second initial belief of empirical event E, P([at ,v t ,j t ]|N) represents the inverse conditional probability model of routine event N, and P(N) represents the second initial belief of routine event N.
[0041] Preferably, the global iterative belief is determined based on the iterative belief, and the specific process is as follows:
[0042] Based on a comprehensive judgment of the iterative beliefs of the aggressive event A, the cautious event C, the empirical event E, and the routine event N, the iterative belief of the event with the highest probability index is selected as the global iterative belief.
[0043] The belief estimation and rolling iteration system for hybrid traffic cognitive uncertainty includes: an information acquisition module, a subjective hypothesis module, an actual observation module, an iterative correction module, and a loop execution module;
[0044] The information acquisition module is used to acquire motion data of participants in mixed traffic.
[0045] The subjective hypothesis module is used to determine a first initial belief about cognitive uncertainty and multiple types of events, and randomly distribute the initial belief to the multiple types of events to obtain a second initial belief for each type of event;
[0046] The actual observation module is used to determine an objective belief about the mixed traffic situation based on the motion data and the multiple types of events.
[0047] The iterative correction module is used to divide the state variables into intervals and match them with the multiple types of events to obtain the inverse conditional probability model for each type of event; it is also used to obtain the iterative belief for each type of event based on the second initial belief, the objective belief and the inverse conditional probability model, and determine the global iterative belief as the cognitive result output based on the iterative belief;
[0048] The loop execution module is used to repeatedly execute the above process, assigning the current iterative belief to the initial belief distributed to each type of event in the next loop stage, until the final cognitive result meets the preset value.
[0049] As can be seen from the above technical solutions, compared with the prior art, the present invention discloses a method and system for belief estimation and rolling iteration of cognitive uncertainty in mixed traffic. By incorporating the difficult-to-handle random uncertainty problem into the design process of intelligent vehicle situational awareness system, and realizing the spontaneous construction and rolling iteration of cognitive understanding, it helps to enhance the intelligent vehicle's cognition and understanding of mixed traffic environment, giving the intelligent vehicle the advantage of human-like cognitive logic; at the same time, it also helps to improve the intelligence level and anthropomorphism of intelligent vehicles, as well as the adaptability and integration of intelligent vehicles to the environment, and helps to improve the speed, accuracy and robustness of intelligent vehicle situational awareness, laying the foundation for intelligent vehicles to truly and naturally integrate into the mixed traffic ecosystem, which is of great significance to the industrialization of intelligent vehicle technology. Attached Figure Description
[0050] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.
[0051] Figure 1 A flowchart of the belief estimation and rolling iteration method for hybrid traffic cognitive uncertainty provided by the present invention.
[0052] Figure 2 A flowchart of the belief estimation and rolling iteration method for hybrid traffic cognitive uncertainty provided by the present invention.
[0053] Figure 3 This invention provides a schematic diagram of the mapping between four types of events and three observation aspects.
[0054] Figure 4a The initial setting provided by this invention is event type A, and the schematic diagram shows the result of the intelligent car's cognitive belief change for vehicle 1 in week 1.
[0055] Figure 4b The initial setting provided by this invention is event type A, and the schematic diagram shows the result of the intelligent car's cognitive belief change regarding car 2.
[0056] Figure 4c The initial setting provided by this invention is event type A, and the schematic diagram shows the result of the intelligent vehicle's cognitive belief change regarding vehicle 3.
[0057] Figure 4d The initial setting provided by this invention is event type A, and the schematic diagram shows the result of the intelligent vehicle's cognitive belief change regarding vehicle 4.
[0058] Figure 5aThe initial setting provided by this invention is event type C, and the schematic diagram shows the result of the intelligent vehicle's cognitive belief change for vehicle 1 in week 1.
[0059] Figure 5b The initial setting provided by this invention is event type C, and the schematic diagram shows the result of the intelligent vehicle's cognitive belief change for car 2 in week 2.
[0060] Figure 5c The initial setting provided by this invention is event type C, and the schematic diagram shows the result of the intelligent vehicle's cognitive belief change for vehicle 3 in week 3.
[0061] Figure 5d The initial setting provided by this invention is event type C, and the schematic diagram shows the result of the intelligent vehicle's cognitive belief change for vehicle 4.
[0062] Figure 6 A schematic diagram of the belief estimation and rolling iteration system for hybrid traffic cognitive uncertainty provided by the present invention.
[0063] Figure 7 A schematic diagram of the cognitive process of the belief estimation and rolling iteration system for hybrid traffic cognitive uncertainty provided by the present invention. Detailed Implementation
[0064] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0065] Example 1
[0066] like Figure 1 As shown, embodiments of the present invention disclose a method for belief estimation and rolling iteration based on hybrid traffic cognitive uncertainty, including:
[0067] Acquire motion data of participants in mixed traffic;
[0068] Determine the first initial belief about cognitive uncertainty and multiple types of events, and randomly distribute the initial belief to the multiple types of events to obtain the second initial belief for each type of event;
[0069] Objective beliefs about mixed traffic situations are determined based on motion data and multiple types of events;
[0070] The state variables are divided into intervals and matched with multiple types of events to obtain the inverse conditional probability model for each type of event.
[0071] The iterative belief for each type of event is obtained based on the second initial belief, objective belief, and the inverse conditional probability model. The global iterative belief is determined as the cognitive result output based on the iterative belief.
[0072] The above process is repeated, assigning the current iteration belief to the initial belief distributed to each type of event in the next iteration phase, until the final cognitive result meets the preset value.
[0073] Example 2
[0074] like Figure 2 As shown, embodiments of the present invention disclose a method for belief estimation and rolling iteration based on hybrid traffic cognitive uncertainty, including:
[0075] Acquire motion data of participants in mixed traffic.
[0076] Preferably, the process of acquiring motion data is as follows:
[0077] The motion information of traffic participants in mixed traffic is collected and recorded by intelligent vehicle-mounted environmental perception and communication devices; motion data is obtained by direct acquisition or indirect calculations such as integration and differentiation based on the motion information.
[0078] Preferably, the vehicle-mounted environmental perception equipment and communication equipment include at least: lidar, millimeter-wave radar, ultrasonic radar, vision sensor, vehicle-to-vehicle communication device, vehicle-to-infrastructure communication device and vehicle-to-cloud communication device.
[0079] Preferably, the motion data includes at least: instantaneous acceleration, instantaneous velocity, and instantaneous jerk.
[0080] State variables include at least the following: continuous acceleration range, continuous velocity range, and continuous jerk range. State variables are inherent properties in vehicle kinematics.
[0081] Determine the first initial belief about cognitive uncertainty and multiple types of events, and randomly distribute the initial belief to the multiple types of events to obtain the second initial belief for each type of event.
[0082] Preferably, the first initial belief of intelligent vehicles in the uncertainty of mixed traffic cognition aims to establish subjective guesses about multiple types of events based on the intelligent vehicle's previous experience, situational understanding and knowledge reserves before the occurrence of multiple types of events. This fully combines the brain-like subjectivity of intelligent vehicle cognition and reflects the differences in the distribution of personalized cognition of intelligent vehicles.
[0083] Preferably, in this embodiment, at least four types of events are classified and determined based on the style types of traffic participants in mixed traffic: aggressive events A, cautious events C, experiential events E, and routine events N. The first initial belief is randomly distributed to the four types of events based on the situational understanding and knowledge reserves of the intelligent vehicle ontology.
[0084] Preferably, in this embodiment, the four types of events are denoted as Π, and any event is denoted as π. ω , π ω ∈ Π, Π=[A,C,E,N], the second initial belief for any event is P(π) ω The second initial belief for aggressive event A is P(A), the second initial belief for cautious event C is P(C), the second initial belief for experiential event E is P(E), and the second initial belief for routine event N is P(N).
[0085] Preferably, the initial belief about cognitive uncertainty is initially determined as the first initial belief. After the first initial belief is randomly distributed to multiple types of events, the corresponding initial belief for each type of event is obtained as the second initial belief.
[0086] Preferably, in the early stages of forming a knowledge base for mixed traffic situations, intelligent vehicles may lack prior experience, have a shallow understanding of the situation, and possess limited knowledge reserves. This embodiment distributes the intelligent vehicle's initial beliefs about the four types of events evenly as follows: That is, in the initial beliefs about intelligent vehicles, there are The probability is considered an aggressive event, and there is The probability is considered a cautious event, and there is The probability is considered an empirical event, and there is The probability is considered a regular event.
[0087] Preferably, any inconsistency in the distribution of the first initial belief caused by mixed traffic random uncertainties can be considered as a subsequent iterative process of the balanced distribution of the first initial belief in this invention. This invention is adaptable to distributing the initial belief in any way.
[0088] Objective beliefs about mixed traffic situations are determined based on motion data and multiple events.
[0089] The preferred process for determining objective beliefs is as follows:
[0090] Based on instantaneous acceleration a t Instantaneous velocity v t and instantaneous speed j t We conducted practical parallel observations from three perspectives to determine objective beliefs P(O) for multiple types of events:
[0091]
[0092] Where, π ω N represents any event from multiple event classes. π Represents event π ω The quantity, P(π) ω) represents any event π ω The initial belief, P([a t ,v t ,j t ]|π ω ) represents any event π ω The inverse conditional probability model.
[0093] Preferably, the objective belief under the condition that aggressive event A occurs is P(A)P([a]). t ,v t ,j t ]|A), The objective belief under the condition that the cautious event C occurs is P(C)P([a t ,v t ,j t ]|C). The objective belief under the condition that the empirical event E occurs is P(E)P([a t ,v t ,j t ]|E), the objective belief under the condition of the occurrence of routine event N is P(N)P([a t ,v t ,j t ]|N);
[0094] P(O)=P(A)P([a t ,v t ,j t ]|A)+P(C)P([a t ,v t ,j t ]|C)+P(E)P([a t ,v t ,j t ]|E)+P(N)P([a t ,v t ,j t ]|N);
[0095] P([a t ,v t ,j t ]|π ω )=P(a t |π ω )P(v t |π ω )P(j t |π ω );
[0096] Wherein, P(a t |π ω ) represents any event π ω Under the conditions of occurrence, from instantaneous acceleration a tThe conditional probability of making an actual observation, P(v t |π ω ) represents any event π ω Under the conditions of occurrence, from the instantaneous velocity v t The conditional probability of making the actual observation, P(j t |π ω ) represents any event π ω Under the condition of occurrence, from the instantaneous jerk j t The conditional probability of making actual observations.
[0097] The state variables are divided into intervals and matched with multiple types of events to obtain the inverse conditional probability model for each type of event.
[0098] Preferably, the state variables are divided into intervals, and the specific process is as follows:
[0099] The continuous acceleration interval is divided into 0 and Divide into points;
[0100] The continuous velocity range is determined according to v init and Divide into points;
[0101] Continuous rapidity intervals are divided into 0 and Divide into points;
[0102] Among them, a max Indicates the maximum acceleration boundary, a min Denotes the minimum acceleration boundary, v init v represents the initial velocity of traffic participants observed at the start of the first round of belief estimation cycle. max Represents the maximum velocity boundary, v min Denotes the minimum velocity boundary, j max Represents the maximum jerk boundary, j min This indicates the minimum jerkness boundary.
[0103] Preferred, such as Figure 3 As shown, the specific process of matching multiple types of events is as follows:
[0104] Aggressive event A typically exhibits frequent acceleration / deceleration and sustained aggressive driving behavior, and is therefore matched to [0, a]. max Acceleration sub-interval, [v init ,v max Velocity subintervals and [0,j] max ] Rapidity sub-interval;
[0105] Prioritizing driving safety in caution-related events (C) typically manifests as compromising and yielding driving behaviors; therefore, it is matched to [a]min [0] acceleration sub-interval, [v min ,v init ] velocity subinterval and [j min [0] abrupt change sub-interval;
[0106] Experience-based events (E) demonstrate a strong ability to control driving risks, typically balancing high traffic efficiency with a comfortable driving experience; therefore, they are matched to... Acceleration sub-interval, velocity sub-intervals and Rapidity sub-interval;
[0107] The similar likelihood of the presentation of routine events N across various driving behaviors represents a trade-off between aggressive, cautious, and experiential events, uniformly matching them to [a] min ,a max Acceleration across the entire range, [v] min ,v max ] velocity over the entire range and [j min ,j max The entire range of urgency.
[0108] Preferably, within the aforementioned specific matching interval, the intelligent vehicle presents the corresponding event type with a certain probability, and also presents other types besides this type with a certain probability. This invention designs a probabilistic cognitive method that can flexibly cope with the random uncertainty of mixed traffic cognition, prevent extreme cognitive thinking and facilitate misidentification and correction, and can also achieve the accuracy of cognitive belief estimation and the speed of cognitive convergence by adjusting specific probability factors.
[0109] Preferably, the reverse inference conditional probability model for each type of event specifically includes: the reverse inference conditional probability model for aggressive event A, the reverse inference conditional probability model for cautious event C, the reverse inference conditional probability model for empirical event E, and the reverse inference conditional probability model for routine event N.
[0110] Preferably, the inverse conditional probability model P([a] for aggressive event A) is... t ,v t ,j t A) is:
[0111]
[0112] in, This refers to the aggressive event A and the observed aspect a. t Corresponding adjustable probability factors, This refers to the aggressive event A and the observed aspect v. t Corresponding adjustable probability factors, This refers to the aggressive event A and the observed aspect j. t The corresponding adjustable probability factor.
[0113] Preferably, in this embodiment, and All are set to 0.4, indicating that from a t v t and j t Looking at each observation aspect, this behavior leads intelligent vehicles to have a 60% belief that it is event type A, while a 40% belief that it is one of events C, E, and N; this can be adjusted according to their needs by changing the probability factor. and The magnitude of the value can be used to change the implementation effect of the cognitive belief estimation and rolling iteration method established in this invention.
[0114] Preferably, the inverse conditional probability model P([a] for the cautious event C. t ,v t ,j t ]|C) is:
[0115]
[0116] in, This indicates the relationship between the cautious event C and the observation aspect a. t Corresponding adjustable probability factors, Indicates the cautious event C and the observation aspect v t Corresponding adjustable probability factors, This indicates the relationship between the cautious event C and the observed aspect j. t The corresponding adjustable probability factor.
[0117] Preferably, in this embodiment, and All are set to 0.4, indicating that from a t v t and j t Looking at each observation aspect, this behavior leads intelligent vehicles to have a 60% belief that it is event type C, while a 40% belief that it is one of events A, E, and N; this can be adjusted according to their needs by changing the probability factor. and The magnitude of the value can be used to change the implementation effect of the cognitive belief estimation and rolling iteration method established in this invention.
[0118] Preferably, the inverse conditional probability model P([a]) of empirical event E is derived from the backpropagation of the conditional probability model. t ,v t ,j t ]|E) is:
[0119]
[0120] in, Indicates the left division point of the continuous acceleration interval. Indicates the right division point of the continuous acceleration interval. Θ v Indicates the points dividing a continuous velocity interval. Indicates the left division point of the continuous jerk interval. Indicates the right division point of the continuous rapidity interval. This indicates the relationship between empirical events E and observed aspects a. t Corresponding adjustable probability factors, Representing empirical events E and observed aspects v t Corresponding adjustable probability factors, This indicates the relationship between empirical events E and observed aspects j. t The corresponding adjustable probability factor.
[0121] Preferably, in this embodiment, and All are set to 0.4, indicating that from a t v t and j t Looking at each observation aspect, this behavior leads the intelligent vehicle to have a 60% belief that it perceives the event as type E, while also having a 40% belief that it perceives it as one of events A, C, and N; it can adjust the probability factor according to its own needs. and The magnitude of the value can be used to change the implementation effect of the cognitive belief estimation and rolling iteration method established in this invention.
[0122] Preferably, the inverse conditional probability model P([a] for routine events N is used to infer the probability of the event N. t ,v t ,j t ]|N) is:
[0123]
[0124] in, This represents the relationship between routine events N and observed aspects a. t Corresponding adjustable probability factors, Represents the relationship between routine events N and observed aspects v. t Corresponding adjustable probability factors, This represents the relationship between routine events N and observed aspects j. tThe corresponding adjustable probability factor.
[0125] Preferably, in this embodiment, and All set to Indicates from a t v t and j t From each observation aspect, when an observed behavior falls within any of the three sub-intervals of its respective observation aspect, the intelligent vehicle has... The belief is that it is perceived as event type N; it can be adjusted according to one's needs by changing the probability factor. and The magnitude of the value can be used to change the implementation effect of the cognitive belief estimation and rolling iteration method established in this invention.
[0126] The iterative belief for each type of event is obtained based on the second initial belief, objective belief, and the inverse conditional probability model. The global iterative belief is then determined as the cognitive result output based on the iterative belief.
[0127] Preferably, the iterative belief for each type of event is obtained, specifically including:
[0128] Iterative belief P(A|[a] for aggressive event A) t ,v t ,j t ])for:
[0129] Iterative belief P(C|[a] for a cautious event C t ,v t ,j t ])for:
[0130] Iterative belief P(E|[a] for empirical event E t ,v t ,j t ])for:
[0131] Iterative belief P(N|[a] for routine events N t ,v t ,j t ])for:
[0132] Wherein, P([a t ,v t ,j t ]|A) represents the inverse conditional probability model of aggressive event A, P(A) represents the second initial belief of aggressive event A, P([a t ,v t ,j t]|C) represents the inverse conditional probability model of the cautious event C, P(C) represents the second initial belief of the cautious event C, P([a t ,v t ,j t ]|E) represents the inverse conditional probability model of empirical event E, P(E) represents the second initial belief of empirical event E, P([a t ,v t ,j t ]|N) represents the inverse conditional probability model of routine event N, and P(N) represents the second initial belief of routine event N.
[0133] Preferably, the global iterative belief is determined based on the iterative belief, and the specific process is as follows:
[0134] Based on the iterative beliefs of aggressive event A, cautious event C, empirical event E, and routine event N, a comprehensive judgment is made, and the iterative belief of the event with the highest probability index is selected as the global iterative belief. The cognitive results of the intelligent vehicle in the current stage of the belief estimation loop are then output.
[0135] The above process is repeated, assigning the current iteration belief to the initial belief distributed to each type of event in the next iteration phase, until the final cognitive result meets the preset value.
[0136] Preferably, the iterative beliefs of the aggressive event A, the cautious event C, the empirical event E, and the routine event N output by the current stage of the cycle process are continuously assigned to the initial beliefs distributed to the four types of events at the beginning of the next stage of the cycle process in a rolling and iterative manner.
[0137] Preferably, the above process is repeated until the final output cognitive result, i.e. the final output global iterative belief, is greater than or equal to the preset value, and the loop ends. At this time, the output cognitive result serves as the intelligent vehicle's cognition and judgment of the uncertain factors of mixed traffic, and thus accurately determines it as a certain event type.
[0138] Example 3
[0139] To verify the accuracy of the method of this invention, four traffic participants around the intelligent vehicle were first set as event type A, as their essential characteristics. The intelligent vehicle establishes its cognition of the surrounding traffic participants through continuous belief estimation and rolling iteration. The results are as follows: Figures 4a-4dAs shown, during the initial belief estimation loop, the global iterative belief was for event types C and E. However, after multiple rolling iterations, the intelligent vehicle's belief in surrounding traffic participants as event type A gradually converged to 1, while the beliefs for other event types gradually converged to 0. This means the intelligent vehicle gradually deepened its understanding and determined that all surrounding traffic participants were event type A, which aligns with the essential types of the surrounding traffic participants. This consistency with the initially set event types demonstrates the accuracy of the method described in this invention.
[0140] By assigning all four traffic participants surrounding the intelligent vehicle to event type C as its essential characteristic, the intelligent vehicle establishes its cognition of these participants through continuous belief estimation and iterative updates. The results are as follows: Figures 5a-5d As shown, it can be seen that in the belief estimation loop of the first few stages, the global iterative belief is accurately displayed as event type C. After multiple stages of rolling iteration, the intelligent vehicle further deepens its belief that the surrounding traffic participants are event type C, and gradually converges to 1. This is also consistent with the essential type of the surrounding traffic participants. That is, the belief estimation and rolling iteration method for mixed traffic cognitive uncertainty proposed in this invention has high speed and accuracy.
[0141] Example 4
[0142] like Figures 6-7 As shown, the belief estimation and rolling iteration system for hybrid traffic cognitive uncertainty includes: an information acquisition module, a subjective hypothesis module, an actual observation module, an iterative correction module, and a loop execution module;
[0143] The information acquisition module is used to acquire motion data of participants in mixed traffic.
[0144] The subjective hypothesis module is used to determine the first initial belief about cognitive uncertainty and multiple types of events, and randomly distributes the initial belief to multiple types of events to obtain the second initial belief for each type of event;
[0145] The actual observation module is used to determine objective beliefs about mixed traffic situations based on motion data and multiple types of events;
[0146] The iterative correction module is used to divide the state variables into intervals and match them with multiple types of events to obtain the inverse back-inference conditional probability model for each type of event; it is also used to obtain the iterative belief for each type of event based on the second initial belief, objective belief and the inverse back-inference conditional probability model, and to determine the global iterative belief as the cognitive result output based on the iterative belief.
[0147] The loop execution module is used to repeatedly execute the above process, assigning the current iteration belief to the initial belief distributed to each type of event in the next loop stage, until the final cognitive result meets the preset value.
[0148] As can be seen from the above technical solutions, compared with the prior art, the present invention discloses a method and system for belief estimation and rolling iteration of cognitive uncertainty in mixed traffic. By incorporating the difficult-to-handle random uncertainty problem into the design process of intelligent vehicle situational awareness system, and realizing the spontaneous construction and rolling iteration of cognitive understanding, it helps to enhance the intelligent vehicle's cognition and understanding of mixed traffic environment, giving the intelligent vehicle the advantage of human-like cognitive logic; at the same time, it also helps to improve the intelligence level and anthropomorphism of intelligent vehicles, as well as the adaptability and integration of intelligent vehicles to the environment, and helps to improve the speed, accuracy and robustness of intelligent vehicle situational awareness, laying the foundation for intelligent vehicles to truly and naturally integrate into the mixed traffic ecosystem, which is of great significance to the industrialization of intelligent vehicle technology.
[0149] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.
[0150] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A method for estimating and iterating beliefs based on hybrid traffic cognitive uncertainty, characterized in that, include: Acquire motion data of participants in mixed traffic; The motion data includes at least: instantaneous acceleration, instantaneous velocity, and instantaneous jerk. A first initial belief about cognitive uncertainty and multiple types of events are determined, and the first initial belief is randomly distributed to the multiple types of events to obtain a second initial belief for each type of event; The multiple types of events include at least: aggressive events A, cautious events C, experiential events E, and routine events N; Based on the motion data and the various events, an objective belief in the mixed traffic situation is determined; The process of determining the objective belief is as follows: Based on the instantaneous acceleration a t Instantaneous velocity v t and instantaneous speed j t Conduct actual parallel observations and determine objective beliefs for the aforementioned multiple types of events. : ； in, Represents any event among multiple event types. Indicates an event Quantity, Represents any event The initial belief, Represents any event The inverse conditional probability model; The state variables are divided into intervals and matched with the multiple types of events to obtain the inverse conditional probability model for each type of event. Based on the second initial belief, the objective belief, and the reverse inference conditional probability model, the iterative belief for each type of event is obtained, and the global iterative belief is determined as the cognitive result output based on the iterative belief; The iterative beliefs for each type of event are obtained, specifically including: Iterative belief in aggressive event A for: ; Iterative belief in cautious event C for: ; Iterative beliefs of empirical event E for: ; Iterative belief of routine event N for: ; in, This represents the inverse conditional probability model for aggressive event A. The second initial belief represents the aggressive event A. This represents the inverse conditional probability model for a cautious event C. The second initial belief representing the cautious event C, This represents the inverse conditional probability model of empirical event E. This represents the second initial belief regarding empirical event E. This represents the inverse conditional probability model for a routine event N. The second initial belief represents a routine event N; The above process is repeated, assigning the current iterative belief to the initial belief distributed to each type of event in the next iteration stage, until the final cognitive result meets the preset value.
2. The method for belief estimation and rolling iteration based on hybrid traffic cognitive uncertainty according to claim 1, characterized in that, The specific process for acquiring motion data is as follows: The movement information of traffic participants in mixed traffic is collected and recorded through intelligent vehicle-mounted environmental sensing and communication devices. The motion data is obtained based on the motion information.
3. The method for belief estimation and rolling iteration based on hybrid traffic cognitive uncertainty according to claim 2, characterized in that, The state variables include at least: continuous acceleration interval, continuous velocity interval, and continuous jerk interval.
4. The method for belief estimation and rolling iteration based on hybrid traffic cognitive uncertainty according to claim 3, characterized in that, The specific process of dividing the state variable into intervals is as follows: The continuous acceleration interval is divided into 0 and Divide into points; The continuous velocity range is divided into and Divide into points; The continuous jerky intervals are divided into 0 and Divide into points; in, Indicates the maximum acceleration boundary. Indicates the minimum acceleration boundary. This represents the initial speed of traffic participants observed at the start of the first round of belief estimation cycle. Indicates the maximum speed boundary. Indicates the minimum velocity boundary. Indicates the boundary of maximum jerkiness. This indicates the minimum jerkness boundary.
5. The method for belief estimation and rolling iteration based on hybrid traffic cognitive uncertainty according to claim 4, characterized in that, The specific process of matching the aforementioned multiple types of events is as follows: Match the offensive event A to Acceleration sub-interval, velocity sub-intervals and Rapidity sub-interval; Match the cautious event C to Acceleration sub-interval, velocity sub-intervals and Rapidity sub-interval; Match the empirical event E to Acceleration sub-interval, velocity sub-intervals and Rapidity sub-interval; Match the regular event N to Acceleration across the entire range Speed range The entire range of urgency.
6. The belief estimation and rolling iteration method for hybrid traffic cognitive uncertainty according to claim 1, characterized in that, The global iterative belief is determined based on the aforementioned iterative belief, and the specific process is as follows: Based on a comprehensive judgment of the iterative beliefs of the aggressive event A, the cautious event C, the empirical event E, and the routine event N, the iterative belief of the event with the highest probability index is selected as the global iterative belief.
7. A belief estimation and rolling iteration system for hybrid traffic cognitive uncertainty, used to execute the belief estimation and rolling iteration method for hybrid traffic cognitive uncertainty as described in any one of claims 1-6, characterized in that, include: The module includes an information acquisition module, a subjective hypothesis module, an actual observation module, an iterative correction module, and a loop execution module. The information acquisition module is used to acquire motion data of participants in mixed traffic. The subjective hypothesis module is used to determine a first initial belief about cognitive uncertainty and multiple types of events, and randomly distribute the initial belief to the multiple types of events to obtain a second initial belief for each type of event; The actual observation module is used to determine an objective belief about the mixed traffic situation based on the motion data and the multiple types of events. The iterative correction module is used to divide the state variables into intervals and match them with the multiple types of events to obtain the inverse conditional probability model for each type of event; it is also used to obtain the iterative belief for each type of event based on the second initial belief, the objective belief and the inverse conditional probability model, and determine the global iterative belief as the cognitive result output based on the iterative belief; The loop execution module is used to repeatedly execute the above process, assigning the current iterative belief to the initial belief distributed to each type of event in the next loop stage, until the final cognitive result meets the preset value.