Pedestrian implicit cooperative game method and system for passing at unprotected intersection

Abstract

The invention provides a pedestrian implicit cooperative game method and system for passing at an unprotected intersection, and the method comprises the steps: sensing the environment of the unprotected intersection, and obtaining the state information of a pedestrian and the state information of a vehicle; based on a state space, an action space, an observation space and a reward function, the interaction process of the vehicle and the pedestrian is modeled as a partially observable Markov decision process, the state space comprises vehicle state information and pedestrian state information, and the action space comprises a vehicle planning track; the reward function is used for carrying out reward calculation on the collaboration degree of vehicle and pedestrian collaboration; generating a plurality of candidate strategies with intention expressions; for candidate strategies, predicting pedestrian response by using a pedestrian response prediction model, and selecting an optimal strategy based on a reward function; and executing the selected strategy, and performing rolling optimization according to the real reaction of the pedestrian. According to the method, the intelligent driving vehicle can actively and safely carry out implicit negotiation with the pedestrians, and efficient, smooth and safe non-protection intersection passing is realized.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving technology, specifically to a method and system for implicit cooperative game theory among pedestrians crossing unprotected intersections. Background Technology

[0002] In the field of autonomous driving technology, crossing unprotected intersections is a critical and complex application scenario. Its characteristic lies in the need for vehicles and pedestrians to dynamically negotiate right-of-way through interaction in the absence of explicit traffic rules (such as traffic lights). Currently, autonomous driving systems primarily rely on two types of technical solutions when handling unprotected intersection crossings: one is a passive decision-making method based on a prediction-response model, and the other is an intention-transmission method based on explicit communication. While these solutions achieve basic functions to some extent, they fail to effectively simulate the implicit cooperative mechanisms inherent in human driving, limiting their performance and reliability in practical applications. Specifically: First, the prediction-response model suffers from overly conservative and passive flaws. Existing methods typically treat pedestrian behavior as an independent prediction object, employing a step-by-step approach: first, predicting the pedestrian's future trajectory using sensor data, and then planning the vehicle's path based on the prediction results. This method is inherently passive; if the prediction accuracy is insufficient (e.g., due to the multimodal uncertainty of pedestrian behavior), the vehicle is prone to decision-making hesitation, manifesting as sudden braking or prolonged stagnation, leading to intersection "lockdowns." This not only reduces traffic efficiency but may also cause traffic congestion behind, violating the core requirements of smoothness and safety for autonomous driving systems. The fundamental problem lies in failing to treat the vehicle and pedestrians as an interactive whole, instead responding based on isolated predictions, lacking proactive guidance capabilities.

[0003] Second, explicit communication methods have limitations in practicality and universality. While conveying vehicle intentions to pedestrians through external devices (such as displays and light signals) achieves explicit communication, this method relies on pedestrians' accurate interpretation and trust of external signals, facing multiple challenges in practical applications: First, pedestrians may misunderstand non-standardized signals, especially in scenarios with multiple traffic participants, easily leading to information confusion; second, such technologies are not yet widespread and involve legal compliance and standardization issues, making large-scale deployment difficult. Therefore, explicit communication cannot reliably replace the natural implicit interaction in human driving, limiting its applicability in complex urban environments.

[0004] Third, existing technologies lack genuine interactive and collaborative mechanisms. Neither predictive-response models nor explicit communication model vehicles and pedestrians are modeled as collaborative decision-making units. These methods ignore the implicit collaborative process in human driving—the transmission of intent through vehicle movement (such as slowing down and changes in acceleration)—which is crucial for efficient and smooth traffic flow. Existing solutions often simplify interaction to one-way or inefficient command transmission, failing to simulate the tacit negotiation between human drivers. This results in rigid and unpredictable vehicle behavior, impacting the safety and comfort of overall traffic flow.

[0005] In summary, existing technologies are mainly limited by core issues such as passivity, communication dependence, and lack of coordination when dealing with unprotected intersections. There is an urgent need in this field for a technical solution that can simulate the implicit interaction logic of humans, enabling intelligent driving vehicles to make decisions in a proactive and safe manner in collaboration with pedestrians. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for implicit cooperative game theory for pedestrians crossing unprotected intersections, overcoming the technical problem that intelligent driving vehicles in the prior art are passive and inefficient at unprotected intersections.

[0007] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: In a first aspect, the present invention provides an implicit collaborative game method for pedestrians crossing unprotected intersections, comprising the following steps: S1. Use onboard sensors to sense the environment of unprotected intersections and obtain pedestrian and vehicle status information; S2. The interaction process between the vehicle and the pedestrian is modeled as a partially observable Markov decision process based on the state space, action space, observation space, and reward function. The state space includes the state information of the vehicle and the state information of the pedestrian. The action space includes the planned trajectory of the vehicle. The observation space is the local observation space of the pedestrian's actions on the vehicle. The reward function includes a reward calculation for the degree of cooperation between the vehicle and the pedestrian. S3. Based on the partially observable Markov decision process, generate multiple candidate strategies with intent expressions. Each candidate strategy corresponds to a vehicle motion trajectory and is given interactive semantics. S4. For the candidate strategies, the pedestrian response prediction model is used to predict the pedestrian's response, and the optimal strategy is selected based on the reward function. S5. Execute the selected optimal strategy and perform rolling optimization based on the actual reactions of pedestrians.

[0008] Based on the above scheme, the sensor in step S1 further includes lidar, camera or millimeter-wave radar; The pedestrian status information includes location, speed, acceleration, walking direction, and head orientation; The vehicle status information includes position, speed, acceleration, and jerk.

[0009] Furthermore, the reward function in step S2 is a multi-objective function, which includes at least security rewards, traffic efficiency rewards, and cooperation rewards.

[0010] Furthermore, the candidate strategies in step S3 include a prior strategy, a yielding strategy, or a cooperative passage strategy.

[0011] Furthermore, step S4 specifically includes: for each candidate strategy, using a pedestrian response prediction model to predict the pedestrian's response, outputting the probability distribution of the pedestrian's future trajectory based on historical interaction semantic data, and selecting the strategy that maximizes the expected cumulative reward as the optimal execution strategy.

[0012] Furthermore, the pedestrian reaction prediction model is a deep learning network that learns pedestrian behavior patterns through training data and outputs probability distributions including whether a pedestrian stops, accelerates through, or walks normally.

[0013] Furthermore, the rolling optimization in step S5 includes continuously sensing changes in pedestrian status and repeating steps S1 to S4 at fixed time intervals to replan for closed-loop control.

[0014] Secondly, the present invention also provides a pedestrian implicit cooperative game system for unprotected intersection crossings to implement the above method, comprising: The perception module is configured to sense the environment of unprotected intersections, pedestrian status, and vehicle status through sensors. The collaborative game decision-making module includes a modeling unit, a strategy generation unit, a pedestrian response predictor, and a strategy evaluation and selection unit, configured to perform modeling, candidate strategy generation, pedestrian response prediction, and strategy selection. The trajectory planning and control module is configured to convert the optimal strategy trajectory into an executable path and speed, and send it to the vehicle drive-by-wire system. The rolling optimization module is configured to perform rolling optimization for closed-loop control.

[0015] Thirdly, the present invention also provides an electronic device, including at least one processor and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the at least one processor implements the implicit collaborative game method for pedestrian crossing at unprotected intersections as described in the first aspect by executing the instructions stored in the memory.

[0016] Fourthly, the present invention also provides a computer-readable storage medium storing at least one instruction or at least one program, wherein the at least one instruction or at least one program is loaded and executed by a processor to implement the implicit cooperative game method for pedestrian crossing at unprotected intersections as described in the first aspect.

[0017] This invention models the interaction between vehicles and pedestrians as a cooperative game process. By generating vehicle trajectories with explicit "intent expressions," it proactively influences pedestrian decisions, guiding them to reach a tacit understanding with the vehicle. This simulates the tacit interaction logic between human drivers, enabling intelligent driving vehicles to proactively and safely engage in implicit negotiation with pedestrians, thereby achieving efficient, smooth, and safe passage through unprotected intersections. Compared with existing technologies, this invention has the following advantages: 1. A paradigm shift from passive to active interaction: This invention transforms autonomous vehicles from passive "observer-responder" to active "interaction participant," proactively guiding the scene in a favorable direction by generating trajectories with clear intentions, thus completely solving the "blocked" problem at intersections.

[0018] 2. Efficient and smooth traffic experience: Through implicit collaboration, vehicles and pedestrians can quickly reach a tacit understanding on the road, avoiding unnecessary complete stops and starts, and significantly improving the traffic efficiency and passenger comfort at intersections.

[0019] 3. Enhanced anthropomorphism and predictability: The driving behavior generated by this invention is more in line with human social interaction habits, making it easier for pedestrians to understand the vehicle's intentions, thereby reducing misunderstandings, enhancing trust, and improving overall traffic safety.

[0020] 4. Strong uncertainty handling capability: Based on the POMDP framework and pedestrian response prediction model, the system can fully consider the multimodal uncertainty of pedestrian behavior and make more robust decisions, rather than relying on a single prediction result that may be wrong.

[0021] In summary, this invention models the interaction between vehicles and pedestrians as a partially observable Markov decision process and generates trajectories with intentional expression, achieving a paradigm shift from passive prediction and response to active collaborative game theory. This improves the efficiency and smoothness of traffic at unprotected intersections, avoids vehicle "lockdowns" and sudden braking, thereby reducing traffic congestion and enhancing passenger comfort. The implicit collaborative mechanism enhances the anthropomorphism and predictability of behavior, making it easier for pedestrians to understand vehicle intentions, reducing misunderstandings and improving overall safety. Finally, based on deep learning prediction, the system can robustly handle the multimodal uncertainty of pedestrian behavior, ensuring the adaptability and reliability of decision-making, ultimately achieving an efficient, safe, and natural experience of navigating unprotected intersections. Attached Figure Description

[0022] Figure 1 A schematic diagram of an implicit collaborative game method for pedestrians crossing unprotected intersections provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the intersection U-turn and maneuvering system based on an advanced urban driver assistance system provided in an embodiment of the present invention; Figure 3 A schematic diagram of the hardware structure of a possible electronic device provided in an embodiment of the present invention; Figure 4 A schematic diagram of the hardware structure of a possible computer-readable storage medium provided for an embodiment of the present invention; Figure 5 A comparison diagram of static path planning and dynamic path planning provided for embodiments of the present invention. Detailed Implementation

[0023] The implementation methods of this solution will be described in further detail below. Obviously, the described embodiments are only a part of the embodiments of this solution, and not an exhaustive list of all embodiments. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of this solution can be combined with each other.

[0024] Example 1 This invention provides an implicit collaborative game method for pedestrians crossing unprotected intersections, such as... Figure 1 As shown, it includes the following steps: Step S1: Use onboard sensors to sense the environment of the unprotected intersection and obtain pedestrian status information and vehicle status information.

[0025] Specifically, the vehicle uses onboard sensors to perceive the environment of unprotected intersections, identifies at least one key pedestrian and obtains pedestrian status information, including position, speed, acceleration, walking direction and head orientation; at the same time, it obtains vehicle status information based on sensors, including position, speed, acceleration and jerk.

[0026] Sensors include lidar, cameras, or millimeter-wave radar.

[0027] This step provides accurate and real-time environmental situation input for subsequent decision-making, overcoming prediction errors caused by incomplete or insufficient perception information at the source, and is a key data foundation for achieving reliable collaborative decision-making.

[0028] Step S2: Based on the state space, action space, observation space, and reward function, the interaction process between the vehicle and the pedestrian is modeled as a partially observable Markov decision process.

[0029] The state space includes vehicle state information and pedestrian state information, the action space includes the planned trajectory of the vehicle, the observation space is the local observation space of the pedestrian's actions on the vehicle, and the reward function includes a reward calculation for the degree of cooperation between the vehicle and the pedestrian.

[0030] Specifically, the reward function is a multi-objective function, including at least safety rewards, traffic efficiency rewards, and cooperation rewards. Cooperation rewards are used to encourage vehicles to produce behaviors that clearly convey intent and guide pedestrians to make the desired responses.

[0031] This step transforms the traditional one-way "prediction-response" model into a two-way "cooperation-game" model. By defining the observation space, it explicitly acknowledges the local dependence of pedestrian decisions on vehicle actions. In particular, the introduced cooperation degree reward function incentivizes vehicles to generate behaviors that proactively and clearly convey intentions at the algorithmic level, providing a core mathematical framework and optimization objective for achieving anthropomorphic implicit cooperation.

[0032] Step S3: Generate multiple candidate strategies with intent expressions based on the partially observable Markov decision process (POMDP model). Each candidate strategy corresponds to a vehicle motion trajectory and is assigned interactive semantics.

[0033] Specifically, Figure 5 The diagram shows a comparison between static and dynamic path planning, with each strategy corresponding to a future trajectory. Each trajectory not only satisfies dynamic constraints but is also assigned different "interaction semantics," with candidate strategies including advance strategies, yielding strategies, or cooperative passage strategies. For example: Preceding strategy: Generate a smooth, steady, and predictable trajectory of constant or slow acceleration, conveying the intention of "I will pass".

[0034] The yielding strategy: Create a clear slowing or stopping trajectory to clearly convey the intention of "please go first" to pedestrians.

[0035] Cooperative crossing strategy: Creates a subtle deceleration and path offset to make room for pedestrians and invite them to cross in sync.

[0036] This step transforms abstract collaborative intentions into concrete, feasible motion trajectories that conform to vehicle dynamics. Each trajectory is a complete "subtext" signal. For example, a steady, uniform trajectory conveys the intention of "going first," while a significantly decelerated trajectory conveys the intention of "yielding." This provides vehicles with diverse and interpretable interaction options and is a key step in the transformation of intentions from "concept" to "execution."

[0037] Step S4: For candidate strategies, use the pedestrian response prediction model to predict the pedestrian's response and select the optimal strategy based on the reward function.

[0038] Specifically, for each candidate strategy, the pedestrian response prediction model is used to predict the pedestrian's response, the probability distribution of the pedestrian's future trajectory is output based on historical interaction semantic data, and the strategy that maximizes the expected cumulative reward is selected as the optimal execution strategy.

[0039] The pedestrian reaction prediction model described above is a deep learning network that learns pedestrian behavior patterns through training data and outputs probability distributions including whether a pedestrian stops, speeds up, or walks normally.

[0040] More specifically, for each candidate strategy, the system uses a trained pedestrian response prediction model to predict several possible responses (such as stopping, speeding through, or walking normally) that pedestrians are most likely to take under that strategy trajectory. Finally, the strategy that maximizes the expected cumulative reward is selected as the execution strategy.

[0041] This step solves the core uncertainty problem in interaction. By simulating the possible responses of users under different strategies, it can proactively evaluate the interaction effects of strategies, thereby actively selecting the strategy most likely to induce safe and efficient collaborative results. This transforms decision-making from "guessing" to "closed-loop simulation optimization," greatly improving the intelligence and success rate of decision-making.

[0042] Step S5: Execute the selected optimal strategy and perform rolling optimization based on the actual reactions of pedestrians.

[0043] The rolling optimization includes continuously sensing changes in pedestrian status and repeating steps S1 to S4 at fixed time intervals to replan for closed-loop control.

[0044] Specifically, the initial part of the selected strategy is executed while continuously sensing the pedestrian's actual reactions. Based on the new state information, steps S1 to S4 are repeated to perform rolling time-domain optimization, dynamically adjusting the strategy to respond to changes in pedestrian behavior, thus forming a closed-loop control.

[0045] Building upon the steps described above, this step enables closed-loop, adaptive control of the dynamic interaction process. When the actual behavior of pedestrians does not match the prediction, the system can adjust its strategy in real time, thereby effectively responding to emergencies, ensuring the robustness and safety of the collaborative process, avoiding intersection standstill or danger caused by a one-time planning error, and ultimately achieving a smooth, seamless, and safe passage experience.

[0046] Example 2 This embodiment provides an intersection U-turn and maneuvering system based on an advanced urban driver assistance system for implementing the method of Embodiment 1 described above, such as... Figure 2 As shown, the system includes: The perception module is configured to sense the environment of unprotected intersections, pedestrian status, and vehicle status through sensors. The collaborative game decision-making module includes a modeling unit, a strategy generation unit, a pedestrian response predictor, and a strategy evaluation and selection unit, configured to perform modeling, candidate strategy generation, pedestrian response prediction, and strategy selection. The trajectory planning and control module is configured to convert the optimal strategy trajectory into an executable path and speed, and send it to the vehicle drive-by-wire system. The rolling optimization module is configured to perform rolling optimization for closed-loop control.

[0047] This system corresponds to the aforementioned turning and maneuvering method, and therefore its modules have the technical effects corresponding to the corresponding method, which will not be elaborated here.

[0048] Example 3 See Figure 3 , Figure 3 This is a schematic diagram illustrating an embodiment of the electronic device provided in this invention. For example... Figure 3 As shown, this embodiment of the invention provides an electronic device 300, including a memory 310, a processor 320, and a computer program 311 stored in the memory 310 and executable on the processor 320. When the processor 320 executes the computer program 311, it performs the following steps: S1. Use onboard sensors to sense the environment of unprotected intersections and obtain pedestrian and vehicle status information; S2. The interaction process between the vehicle and the pedestrian is modeled as a partially observable Markov decision process based on the state space, action space, observation space and reward function. The state space includes the state information of the vehicle and the state information of the pedestrian. The action space includes the planned trajectory of the vehicle. The observation space is the local observation space of the pedestrian on the vehicle's actions. The reward function includes the reward calculation for the degree of cooperation between the vehicle and the pedestrian. S3. Generate multiple candidate strategies with intent expressions based on partially observable Markov decision processes. Each candidate strategy corresponds to a vehicle motion trajectory and is given interactive semantics. S4. For candidate strategies, use the pedestrian response prediction model to predict the pedestrian's response and select the optimal strategy based on the reward function. S5. Execute the selected optimal strategy and perform rolling optimization based on the actual reactions of pedestrians.

[0049] Example 4 See Figure 4 , Figure 4 This is a schematic diagram illustrating an embodiment of a computer-readable storage medium provided by the present invention. (See diagram below.) Figure 4 As shown, this embodiment provides a computer-readable storage medium 400 on which a computer program 311 is stored. When the computer program 311 is executed by a processor, it performs the following steps: S1. Use onboard sensors to sense the environment of unprotected intersections and obtain pedestrian and vehicle status information; S2. The interaction process between the vehicle and the pedestrian is modeled as a partially observable Markov decision process based on the state space, action space, observation space and reward function. The state space includes the state information of the vehicle and the state information of the pedestrian. The action space includes the planned trajectory of the vehicle. The observation space is the local observation space of the pedestrian on the vehicle's actions. The reward function includes the reward calculation for the degree of cooperation between the vehicle and the pedestrian. S3. Generate multiple candidate strategies with intent expressions based on partially observable Markov decision processes. Each candidate strategy corresponds to a vehicle motion trajectory and is given interactive semantics. S4. For candidate strategies, use the pedestrian response prediction model to predict the pedestrian's response and select the optimal strategy based on the reward function. S5. Execute the selected optimal strategy and perform rolling optimization based on the actual reactions of pedestrians.

[0050] This invention models the interaction between vehicles and pedestrians as a cooperative game process. By generating vehicle trajectories with explicit "intent expressions," it proactively influences pedestrian decisions, guiding them to reach a tacit understanding with the vehicle. This simulates the tacit interaction logic between human drivers, enabling intelligent driving vehicles to proactively and safely engage in implicit negotiation with pedestrians, thereby achieving efficient, smooth, and safe passage through unprotected intersections. Compared with existing technologies, this invention has the following advantages: 1. A paradigm shift from passive to active interaction: This invention transforms autonomous vehicles from passive "observer-responder" to active "interaction participant," proactively guiding the scene in a favorable direction by generating trajectories with clear intentions, thus completely solving the "blocked" problem at intersections.

[0051] 2. Efficient and smooth traffic experience: Through implicit collaboration, vehicles and pedestrians can quickly reach a tacit understanding on the road, avoiding unnecessary complete stops and starts, and significantly improving the traffic efficiency and passenger comfort at intersections.

[0052] 3. Enhanced anthropomorphism and predictability: The driving behavior generated by this invention is more in line with human social interaction habits, making it easier for pedestrians to understand the vehicle's intentions, thereby reducing misunderstandings, enhancing trust, and improving overall traffic safety.

[0053] 4. Strong uncertainty handling capability: Based on the POMDP framework and pedestrian response prediction model, the system can fully consider the multimodal uncertainty of pedestrian behavior and make more robust decisions, rather than relying on a single prediction result that may be wrong.

[0054] In summary, this invention models the interaction between vehicles and pedestrians as a partially observable Markov decision process and generates trajectories with intentional expression, achieving a paradigm shift from passive prediction and response to active collaborative game theory. This improves the efficiency and smoothness of traffic at unprotected intersections, avoids vehicle "lockdowns" and sudden braking, thereby reducing traffic congestion and enhancing passenger comfort. The implicit collaborative mechanism enhances the anthropomorphism and predictability of behavior, making it easier for pedestrians to understand vehicle intentions, reducing misunderstandings and improving overall safety. Finally, based on deep learning prediction, the system can robustly handle the multimodal uncertainty of pedestrian behavior, ensuring the adaptability and reliability of decision-making, ultimately achieving an efficient, safe, and natural experience of navigating unprotected intersections.

[0055] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for implicit collaborative game theory among pedestrians crossing unprotected intersections, characterized in that, Includes the following steps: S1. Use onboard sensors to sense the environment of unprotected intersections and obtain pedestrian and vehicle status information; S2. The interaction process between the vehicle and the pedestrian is modeled as a partially observable Markov decision process based on the state space, action space, observation space, and reward function. The state space includes the state information of the vehicle and the state information of the pedestrian. The action space includes the planned trajectory of the vehicle. The observation space is the local observation space of the pedestrian's actions on the vehicle. The reward function includes a reward calculation for the degree of cooperation between the vehicle and the pedestrian. S3. Based on the partially observable Markov decision process, generate multiple candidate strategies with intent expressions. Each candidate strategy corresponds to a vehicle motion trajectory and is given interactive semantics. S4. For the candidate strategies, the pedestrian response prediction model is used to predict the pedestrian's response, and the optimal strategy is selected based on the reward function. S5. Execute the selected optimal strategy and perform rolling optimization based on the actual reactions of pedestrians.

2. The method according to claim 1, characterized in that, The sensors mentioned in step S1 include lidar, cameras, or millimeter-wave radar; The pedestrian status information includes location, speed, acceleration, walking direction, and head orientation; The vehicle status information includes position, speed, acceleration, and jerk.

3. The method according to claim 1, characterized in that, The reward function in step S2 is a multi-objective function, which includes at least security rewards, traffic efficiency rewards, and cooperation rewards.

4. The method according to claim 1, characterized in that, The candidate strategies in step S3 include a prior strategy, a yielding strategy, or a cooperative approach strategy.

5. The method according to claim 1, characterized in that, Step S4 specifically includes: for each candidate strategy, using a pedestrian response prediction model to predict the pedestrian's response, outputting the probability distribution of the pedestrian's future trajectory based on historical interaction semantic data, and selecting the strategy that maximizes the expected cumulative reward as the optimal execution strategy.

6. The method according to claim 5, characterized in that, The pedestrian reaction prediction model is a deep learning network that learns pedestrian behavior patterns through training data and outputs a probability distribution including whether a pedestrian stops, speeds up, or walks normally.

7. The method according to claim 1, characterized in that, The rolling optimization in step S5 includes continuously sensing changes in pedestrian status and repeating steps S1 to S4 at fixed time intervals to replan for closed-loop control.

8. A pedestrian implicit cooperative game system for crossing unprotected intersections, used to implement the method of any one of claims 1 to 7, characterized in that, include: The perception module is configured to sense the environment of unprotected intersections, pedestrian status, and vehicle status through sensors. The collaborative game decision-making module includes a modeling unit, a strategy generation unit, a pedestrian response predictor, and a strategy evaluation and selection unit, configured to perform modeling, candidate strategy generation, pedestrian response prediction, and strategy selection. The trajectory planning and control module is configured to convert the optimal strategy trajectory into an executable path and speed, and send it to the vehicle drive-by-wire system. The rolling optimization module is configured to perform rolling optimization for closed-loop control.

9. An electronic device, comprising at least one processor and a memory communicatively connected to said at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, which implements the pedestrian implicit cooperative game method for unprotected intersection crossing as described in any one of claims 1 to 7 by executing the instructions stored in the memory.

10. A computer-readable storage medium storing at least one instruction or at least one program, said at least one instruction or at least one program being loaded and executed by a processor to implement the pedestrian implicit cooperative game method for unprotected intersection crossing as described in any one of claims 1 to 7.

Images