CN-122024499-A - Traffic control method for signalless intersection based on vehicle cooperation and gap self-adaption

CN122024499ACN 122024499 ACN122024499 ACN 122024499ACN-122024499-A

Abstract

The invention discloses a traffic control method of a signalless intersection based on vehicle cooperation and gap adaptation, which comprises the steps of S1, adopting a layered cooperation mixed traffic management architecture to decouple and cooperate CAV and HDV interaction, simultaneously decoupling and cooperating CAV and pedestrian interaction, S2, providing a CAV-HDV joint decision planning model integrating fuzzy logic risk perception and game theory, converting the dynamic state of an interactive vehicle into quantitative evaluation of human driver perception risk, tightly coupling discrete game output with a continuous MILP-MPC motion planner, S3, setting up a cooperative control mechanism which is actively initiated by the CAV and creates a traversable gap for a pedestrian, and converting a pedestrian gap receiving model from a passive evaluation tool into active input of the CAV motion planning. The invention effectively stabilizes the fluctuation caused by human behavior uncertainty and fully releases the potential of CAV in the mixed traffic environment.

Inventors

WANG XUE
Cao ningbo
ZHAO LIYING
GUO JING
CHENG YAXIAN
WANG JIAMING

Assignees

宁夏理工学院

Dates

Publication Date: 20260512
Application Date: 20260121

Claims (10)

1. A traffic control method for a signalless intersection based on vehicle cooperation and clearance adaptation is characterized by comprising the following steps: S1, adopting a layered collaborative hybrid traffic management architecture to decouple and collaborate interaction of CAV and HDV, and defining an interaction mode of peer-to-peer interaction based on game theory between the CAV and the HDV; S2, providing a CAV-HDV combined decision planning model fused with fuzzy logic risk perception and game theory, converting the dynamic state of an interactive vehicle into quantitative evaluation of human driver perceived risk, and tightly coupling discrete game output with a continuous MILP-MPC motion planner; S3, setting up a cooperative control mechanism which is actively initiated by CAV and creates a traversable gap for pedestrians, converting a pedestrian gap receiving model from a passive evaluation tool into active input of CAV motion planning, and constructing a gap adaptation vehicle group to perform cooperative deceleration.
2. The traffic control method of the signalless intersection based on vehicle coordination and gap adaptation as claimed in claim 1, wherein the step S1 is characterized in that the traffic control method of the signalless intersection is implemented by dividing a spatial structure into a close area, a coordination control area and a collision area, realizing real-time information sharing and coordination planning of position, speed and acceleration between CAVs, simultaneously, CAVs obtains dynamic information of HDVs and pedestrians through detection of an on-board sensor, CAVs is uniformly coordinated and controlled by a central controller, and HDVs and pedestrians have autonomous decision characteristics.
3. The traffic control method of signalless intersection based on vehicle coordination and gap adaptation of claim 1, wherein step S1 uses a MILP-MPC based CAV coordination motion planning method to model the motion planning problem as a mixed integer linear programming problem and uses model predictive control for rolling optimization.
4. The traffic control method of signalless intersection based on vehicle coordination and gap adaptation of claim 3, wherein step S1 generates global optimized trajectories for all CAVs and performs the following rolling optimization using model predictive control: a) The state acquisition, namely acquiring the current states of all CAVs by a central controller through a communication network at the starting moment of each control period; b) Based on the current state, the central controller solves the mixed integer linear programming problem to obtain the optimal control sequences of all CAVs in a time domain in the future; c) The instruction issuing and executing only issues the first control quantity in each CAV control sequence to the corresponding CAV executing; d) And (3) rolling and advancing, namely repeating the steps a to c until the next control period, and re-optimizing according to the new system state.
5. The traffic control method of signalless intersection based on vehicle coordination and gap adaptation of claim 1, wherein step S2 adopts a fuzzy logic based risk perception model to dynamically quantify the interaction of HDV and CAV into a risk value for predicting the traffic intention of HDV.
6. The signalless intersection traffic control method based on vehicle coordination and gap adaptation of claim 5, wherein the step S2 includes: calculating three key interaction indexes of speed difference of the two parties, distance difference reaching conflict point and time difference reaching conflict point in real time as input values of fuzzy logic; Converting the input value into the membership of the linguistic variable through a predefined membership function; Converting the inferred fuzzy set of outputs into an accurate driver perceived risk value Is a subset of linguistic variables; According to the calculated And HDV driving style, predictive aggressive driver threshold Threshold value of normal driver Threshold value of conservative driver ); The traffic decision prediction process of the HDV in the intersection is that if The HDV is accelerated to pass through if The HDV keeps constant speed, if HDV slows down and lets go.
7. The traffic control method of signalless intersection based on vehicle coordination and gap adaptation of claim 1, wherein the CAV in step S2 is based on the predicted HDV decision result to maximize the comprehensive utility as follows; a) The trust utility function refers to the utility brought by the trust of the HDV to the CAV when the behavior strategies of the two parties reach the reciprocal state, and the calculation formula is as follows: ; ; wherein: And Respectively representing the effect and loss of the achievement and non-achievement of reciprocity of CAV and HDV behavior strategies; And Respectively representing the traffic strategy of the vehicle, wherein the traffic is 1, and the avoidance is 0; is a binary variable, and indicates whether the two parties achieve a reciprocity condition; b) The safety utility function refers to the utility that vehicles can pass through intersections in sequence without collision accidents and psychological threats, and is processed by adopting a piecewise function, and the calculation formula is as follows: ; wherein: For an acceptable safe-passing interval of the vehicle, A threshold value for increasing the safety utility of the vehicle beyond which the safety utility of the vehicle will not increase if CAV suffers from collision safety loss if The running environment of the two vehicles is safe, and positive safety effect can be generated at the moment; c) The efficiency utility function refers to the utility generated by the time when the vehicle passes through the intersection, and the calculation formula is as follows: ; ; In the formula, Representing a traffic policy selection of the CAV, Represents the time required to pass at maximum speed; d) CAV integrated utility function: Will trust utility functions Safety utility function And efficiency utility function Integrating and establishing a CAV comprehensive utility function: ; In the formula, The weight parameters of the security utility, the efficiency utility and the trust utility function are respectively, ; The CAV passing decision objective function is the CAV and HDV passing strategy combination when the comprehensive utility function is selected to be maximum : 。
8. The traffic control method of the signalless intersection based on vehicle cooperation and gap adaptation according to claim 1, wherein the step S3 is to divide the pedestrian crossing mode into single-stage continuous crossing, two-stage crossing and rolling gap crossing, and predict the pedestrian crossing mode by determining the information of the environmental vehicle, calculating the maximum time for the pedestrian crossing the lane, judging the feasibility of the pedestrian to accept the gap based on the maximum crossing time, and finally outputting the crossing feasibility judgment result.
9. The traffic control method for the signalless intersection based on vehicle cooperation and gap adaptation according to claim 1, wherein the step S3 determines the gap adaptation vehicle group as follows: searching for the first 2 vehicles at most in each lane, namely, each adaptive vehicle group contains 4 CAVs at most; searching from the conflict point to the upstream lane by lane, and stopping searching in a certain lane once CAV meeting the condition is found in the lane; If a plurality of pedestrians wait at the same time, the adaptive vehicle group will preferentially serve the pedestrians with the highest probability of accepting the gap, and the gap created by the adaptive vehicle group can be shared by other pedestrians; if the pedestrians in two directions are respectively located at the road edge and the central isolation belt, the cooperative planning of the vehicle group is adapted to meet the crossing requirement of the two-way pedestrians at the same time.
10. The traffic control method of the signalless intersection based on vehicle coordination and clearance adaptation of claim 9, wherein the adaptive vehicle group coordination plan in the step S3 is characterized by determining an active clearance adaptive vehicle group, matching pedestrian flows with crossing requirements with the adaptive vehicles, generating crossing mode combinations and performing clearance coordination planning; The traversing mode combination adopts a decision tree structure and is set as follows: The method comprises the steps of taking a virtual root node as a starting point, adding branches for pedestrian flows in each direction in sequence, wherein the branching rules are as follows, and enumerating from the pedestrian or crowd direction with higher probability of passing through gaps preferentially, wherein the pedestrian in each direction corresponds to a plurality of passing through modes, and each existing leaf node carries out branch expansion according to the number of selectable modes; Traversing the child nodes from the root node, and finally forming a complete decision tree containing four leaf nodes, wherein all possible traversing mode combinations are covered; The method comprises the steps that a group of candidate cross-street schemes are arranged from a root node to each leaf node, and among all cross-way combinations, the cross-way combination with the shortest active gap adaptation execution time is the final decision result.

Description

Traffic control method for signalless intersection based on vehicle cooperation and gap self-adaption Technical Field The invention relates to a crossing traffic control method, in particular to a signal-free crossing traffic control method based on vehicle cooperation and gap self-adaption. Background The signalless intersection is a key conflict point in the urban road network, and the operation efficiency and the safety level of the signalless intersection are highly dependent on autonomous negotiation and real-time decision among different traffic participants. With the rise of intelligent network vehicle (CAVs) technology, cooperative traffic control is realized through vehicle-to-vehicle and vehicle-to-road communication, and a new technical path is brought for fundamentally improving the traffic capacity of the intersection. In an ideal pure CAV traffic environment, centralized optimization methods based on Mixed Integer Linear Programming (MILP), model Predictive Control (MPC), etc. have been demonstrated to significantly reduce vehicle delays. However, the real traffic flow will be in a mixed state of CAVs coexisting with the manually driven vehicle (HDVs) and pedestrians for a long period of time, which brings a serious challenge to the practical application of the cooperative control theory. Currently, research and practice oriented to mixed signalless intersections is mainly faced with a core bottleneck of three dimensions: First, at the system level control level, most advanced CAV cooperative control algorithms build on ideal assumptions that all vehicles are fully controllable. When dealing with HDVs whose behavior is uncertain, these algorithms appear fragile because their decisions cannot be accurately predicted, often forcing too conservative yielding strategies. This not only results in the failure of the efficiency advantage of the CAV itself, but also causes a chain reaction due to frequent parking of the CAV, exacerbating the disturbance of the rearward HDV traffic flow, and creating a new source of congestion, a so-called "real gap" problem. Second, at the inter-vehicle interaction level, while methods represented by gambling theory are widely used to characterize the right of way competition of CAV with HDV, existing models typically only output discrete decision instructions such as "pass" or "yield". There is a "decision-planning disjoint" between such instructions and the underlying motion planner responsible for generating smooth, continuous trajectories. The direct consequence is that the vehicle may generate abrupt behaviors such as rapid acceleration, rapid deceleration and the like, which not only affects the riding comfort, but also causes misjudgment of peripheral HDV drivers due to hard actions, thus burying potential safety hazards. Finally, at the vehicle-pedestrian interaction level, the existing research focuses on improving the prediction precision of pedestrian crossing behaviors by using a machine learning or statistical model. However, this "predictive-reactive" mode is passive in nature, with CAV only passively adjusting its trajectory to avoid based on the predicted outcome. The system lacks a mechanism, allows CAV to be used as an active party, and actively shapes and creates a safe and efficient pedestrian crossing window through self cooperative behavior. This results in a system that fails to translate predictive capability into global efficiency and safety gains, which falls into "passive response dilemma". From the above, it is difficult to consider the system-level traffic efficiency, the decision smoothness of the interaction between vehicles, and the active guarantee of pedestrian safety in the mixed and heterogeneous signalless intersection environment in the prior art. Specific defects or deficiencies are as follows: (1) The existing cooperative control framework has the structural limitation, and is difficult to orchestrate heterogeneous traffic participants. The existing scheme mostly follows a single design thought, or combines CAV, HDV and pedestrian interaction, adopts unified control logic, so that a model is complex and poor in adaptability, or performs isolated optimization on one type of interaction only, and the CAV collaborative planning, the HDV game decision and the pedestrian active guarantee are organically integrated by lacking a top-level framework. This mode results in the system making decisions between the modules to catch each other in the face of concurrent collisions with mixed traffic flows, failing to achieve global optimum, synergistic efficiency is especially low at low permeabilities. (2) The existing interaction model is loosely coupled with the motion planner, resulting in a disjoint decision and execution. Although some research attempts to predict HDV behavior by introducing methods such as game theory, their output is typically only an abstract policy directive (e.g. "pass" or "let-down") and is not embedded in the cont