Conditional autonomous driving takeover method, device and equipment based on human-machine trust

Abstract

The present invention belongs to the technical field of autonomous driving takeover, and in particular relates to a conditional autonomous driving takeover method, device and equipment based on human-machine trust, comprising: collecting driving environment information and the driving behavior of the driver on the driving end to generate a trust level value; collecting the generated trust level values, judging the score interval of the trust level values ​​and generating a trust model; generating a weight distribution strategy based on the trust model, and adjusting the steering wheel input weights of the driving end and the autonomous driving end in real time according to the weight distribution module; detecting the weight distribution situation, judging the takeover conditions, and if the takeover conditions are met, completely releasing the lateral control authority of the autonomous driving end and allocating it to the driving end; while completing the handover of the lateral control authority, releasing the longitudinal control authority and allocating it to the driving end. The present invention provides a method for autonomous driving takeover, ensuring that the driver can take over the driving system gently and smoothly, improving consistency and comfort, and enhancing the driver's driving experience.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving takeover technology, and in particular to a method, device, and equipment for conditional autonomous driving takeover based on human-machine trust. Background Art

[0002] The past decade has witnessed tremendous development in autonomous vehicles, posing new challenges for the collaboration between drivers and autonomous vehicles. As autonomous vehicles gain increasing control, the role of the driver needs further clarification. The Society of Automotive Engineers categorizes autonomous vehicles into six levels: Level 0 (no autonomy), Level 1 (driving assistance), Level 2 (partial autonomy), Level 3 (conditional autonomy), Level 4 (high autonomy), and Level 5 (fully autonomous). Due to current technological limitations and legal requirements, the future will be at Level 3 (conditional autonomy). Conditional autonomy means the vehicle can drive autonomously under specific road, traffic, and weather conditions, but requires a human driver to be ready to take over control at all times. The driver's role is crucial. Even though the vehicle can drive autonomously, they must remain ready to take over control to ensure safe driving. The driver needs to be aware of the road and traffic conditions, be ready to take over control at any time, and be able to react and make decisions appropriately. Taking over control of conditionally autonomous vehicles is one of the most challenging tasks in the industry, as it requires a safe and rapid transition between the driver and the autonomous system. While existing takeover strategies can achieve flexible control transitions, these strategies still primarily focus on returning full control of the vehicle to the driver. Drivers may be relaxed, and a sudden takeover can increase the complexity of their driving operations and fail to ensure safety. Furthermore, current takeover strategies fail to consider the driver's trust level and are therefore unable to accurately assess their ability and intent to take over at a specific moment. This can lead to the autonomous system misjudging the driver's ability to take over and failing to understand their intent in emergencies, potentially causing human-machine conflicts, mode confusion, and ultimately posing a traffic safety hazard. Summary of the Invention

[0003] To address the above issues, the present invention provides a method, device, and apparatus for conditional autonomous driving takeover based on human-machine trust. These methods utilize a trust module and corresponding takeover strategies to address potential safety hazards associated with driver takeover. The present invention provides the following technical solutions: In a first aspect, embodiments of the present invention provide a method for conditional autonomous driving takeover based on human-machine trust, comprising:

[0004] Collect the driver's driving behavior and driving environment information, make judgments based on the driving behavior, and generate a trust level value;

[0005] Collect the generated trust level values, determine the score interval of the trust level values ​​and generate a trust model;

[0006] Generate a weight distribution strategy based on the trust model and adjust the weights in real time according to the weight distribution module;

[0007] Detect the weight distribution and determine the takeover conditions. If the takeover conditions are met, release the lateral control authority and assign the lateral control authority to the driver. After the lateral control authority is transferred, release the longitudinal control authority and assign the longitudinal control authority to the driver.

[0008] Furthermore, the collecting of the driving behavior and driving environment information of the driver on the driving side, and making a judgment based on the driving behavior to generate the trust level value further includes collecting the driving environment information and the driving behavior of the driver on the driving side, and making a judgment based on the driving behavior to generate the trust level value further including:

[0009] Collect the driver's driving behavior and driving environment information;

[0010] By collecting relevant physiological characteristics of the driver's driving behavior and comparing them with the system model;

[0011] defining the driver's driving state according to the comparison result;

[0012] A confidence level value is generated based on the driving status.

[0013] Furthermore, the collecting of generated trust level values, determining the score interval of the trust level values ​​and generating a trust model further includes:

[0014] collecting the generated confidence level values;

[0015] Determine the score interval of the trust level value and generate a preliminary mapping function that conforms to the trust level value;

[0016] By calculating the mapping function, a trust model adapted to the current driving status is generated.

[0017] Furthermore, the mapping function used to generate the trust model is:

[0018] g(τ,t)

[0019] Among them, τ is the trust level value, t is the time process, and based on the mapping function g(τ,t), the weight parameter λd(t) used to determine the driver's steering input and the weight parameter λc(t) of the autonomous driving system's steering input are also generated.

[0020] Furthermore, the weight of the driver's steering input λd(t) and the weight of the autonomous driving system's steering input λc(t) satisfy the following conditions:

[0021] λc(t)+λd(t)=1

[0022] That is, the sum of the weight of the driver's steering input and the weight of the autonomous driving system's steering input always sums to 1.

[0023] Furthermore, the generating of the weight distribution strategy according to the trust model and the changing of the weights in real time according to the weight distribution module also include:

[0024] The real-time change trend of the weight is gradually biased towards the driving side, and completely biased towards the driving side at the point preset by the system.

[0025] Furthermore, the detection weight distribution is performed, and the takeover conditions are determined. If the takeover conditions are met, the lateral control authority is released and the lateral control authority is allocated to the driver. After the lateral authority handover is completed, the longitudinal control authority is released and the longitudinal control authority is allocated to the driver. The method further includes:

[0026] δf=λd(t)δd+λc(t)δc

[0027] δf is the steering angle of the vehicle's front wheels, δc is the steering wheel input of the autonomous driving system, and δd is the steering wheel input of the driver. δf is determined collaboratively by the driver and the autonomous driving system. When the value of λd is 1, the vehicle is in autonomous driving mode. When the value of λc is 1, the vehicle is in manual operation mode, and the lateral control authority is released and assigned to the driver to complete the takeover. At the same time, the longitudinal control authority is also released and assigned to the driver to complete the takeover.

[0028] Secondly, this embodiment provides a conditional autonomous driving vehicle takeover system based on human-machine trust, including:

[0029] The trust level generation module is used to collect the driver's driving behavior and driving environment information, and make judgments based on the driving behavior to generate a trust level value;

[0030] A trust model generation module, configured to collect generated trust level values, determine the score interval of the trust level values, and generate a trust model;

[0031] A weight distribution module is used to generate a weight distribution strategy based on the trust model and change the weights in real time according to the weight distribution module;

[0032] The takeover module is used to detect the weight distribution and determine the takeover conditions. If the takeover conditions are met, the lateral control authority is released and allocated to the driver. After the lateral authority handover is completed, the longitudinal control authority is released and allocated to the driver.

[0033] On the third aspect, this embodiment provides a conditional autonomous driving takeover device based on human-machine trust, including a memory, a processor, and a computer program that can run on the processor. When the processor executes the program, it implements any of the conditional autonomous driving vehicle takeover methods based on human-machine trust in this embodiment.

[0034] One or more of the above technical solutions have the following beneficial effects:

[0035] The present invention provides a method for conditional automatic driving takeover based on human-machine trust, which collects the driver's driving behavior based on the trust level generation module and generates an initial trust level value, then collects the driver's trust level value and generates a trust model through the trust model generation module, and then generates a weight distribution strategy based on the generated trust model to realize the system's real-time regulation of the steering wheel weight of the driving end and the steering wheel weight of the automatic driving end, and completes the driver's complete takeover of the vehicle through the takeover module. Among them, the existence of the trust model ensures that the control authority will not be directly transferred to the driving end, ensuring the safety of the driver's takeover, and the existence of the takeover module ensures that the driver on the driving end can completely take over the driving end when conditions are ripe, thereby ensuring the safety of the driver taking over the control authority under conditional automatic driving. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 1 is a flow chart of a conditional autonomous driving takeover method based on human-machine trust provided in an embodiment of the present invention;

[0037] Figure 2 1 is a flow chart of step S1 of a conditional autonomous driving takeover method based on human-machine trust provided in an embodiment of the present invention;

[0038] Figure 3 4 is a flow chart of step S2 of the conditional autonomous driving takeover method based on human-machine trust provided in an embodiment of the present invention;

[0039] Figure 4 is a schematic diagram of a low-level trust model provided in the third embodiment of the present invention;

[0040] Figure 5 is a schematic diagram of a trust model provided in a third embodiment of the present invention that is a medium-level trust model;

[0041] Figure 6is a schematic diagram of a trust model provided in a third embodiment of the present invention that is a high-level trust model;

[0042] Figure 7 It is a flow chart of a conditional autonomous driving takeover system based on human-machine trust provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0043] The embodiments of the present invention are described below with reference to the accompanying drawings.

[0044] In the description of the embodiments of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms, "connection", and "installation" should be understood in a broad sense. For example, "connection" can be a detachable connection or a non-detachable connection; it can be a direct connection or an indirect connection through an intermediate medium. In addition, "communication" can be a direct connection or an indirect connection through an intermediate medium. Here, "fixed" means that the two are connected to each other and the relative position relationship after connection remains unchanged. The directional terms mentioned in the embodiments of the present invention, such as "inside", "outside", "top", "bottom", etc., are only reference to the directions of the accompanying drawings. Therefore, the directional terms used are for better and clearer explanation and understanding of the embodiments of the present invention, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the embodiments of the present invention.

[0045] In the embodiments of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of technical features indicated. Therefore, features defined as "first" or "second" may explicitly or implicitly include one or more of the features.

[0046] In the embodiments of the present invention, "and / or" is simply a description of the association relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone. Furthermore, the character " / " in this document generally indicates that the associated objects are in an "or" relationship.

[0047] References to "one embodiment" or "some embodiments" in this specification mean that a particular feature, structure, or characteristic described in connection with that embodiment is included in one or more embodiments of the present invention. Thus, phrases such as "in one embodiment," "in some embodiments," "in other embodiments," and "in yet other embodiments" appearing in various places in this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "including," "comprising," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0048] Example:

[0049] This embodiment provides a conditional autonomous driving takeover method based on human-machine trust.

[0050] Figure 1 This is a flow chart of the conditional autonomous driving takeover method based on human-machine trust of the present invention. Figures 1 to 3 As shown, an embodiment of the present invention provides a conditional autonomous driving takeover method based on human-machine trust, the method comprising:

[0051] S1. Collecting the driver's driving behavior and driving environment information, and making judgments based on the driving behavior to generate a trust level value;

[0052] S2. Collect the generated trust level values, determine the score interval of the trust level values ​​and generate a trust model;

[0053] S3. Generate a weight distribution strategy based on the trust model and change the weights in real time according to the weight distribution module;

[0054] S4. Detect the weight distribution and determine the takeover conditions. If the takeover conditions are met, release the lateral control authority and assign the lateral control authority to the driver. After the lateral authority handover is completed, release the longitudinal control authority and assign the longitudinal control authority to the driver.

[0055] In an embodiment of the present invention, the driving environment information is first collected to confirm whether a takeover request needs to be initiated. When a takeover request needs to be initiated, the system will collect the driving behavior of the driver on the driving side, make a preliminary judgment, confirm whether it is suitable for takeover and what takeover strategy to adopt, and generate a trust level value containing the driver information and traffic environment information based on the above information; after the trust level value is generated, the trust level value information will be collected and analyzed, and an adaptive trust model will be generated based on the analysis results and the trust level value interval; the system will generate an adaptive weight distribution strategy based on the trust level value, and based on the weight distribution strategy, the system can adjust the direction of the driving side in real time The system checks the weight distribution situation and determines whether it is suitable for the driver to take over completely. When the takeover conditions are met, the system releases the lateral control authority. At this time, the driver will receive the lateral control authority takeover request initiated by the system and complete the takeover of all lateral control authorities. After the takeover is completed, the system releases the longitudinal control authority of the autonomous driving end. At this time, the driver will also receive the longitudinal control authority takeover request initiated by the system and complete the full takeover of the longitudinal control authority.

[0056] The following combination Figure 2 , which details how this embodiment issues a takeover request and collects the driver's driving behavior, and generates a trust level value:

[0057] S11, collecting the driver's driving behavior and driving environment information;

[0058] S12. Obtain relevant physiological characteristics by collecting the driver's driving behavior and compare them with the system model;

[0059] S13. defining the driver's driving state according to the comparison result;

[0060] S14. Generate a confidence level value based on the driving status.

[0061] In this embodiment of the present invention, the system's judgment of the traffic environment is mainly based on the following aspects:

[0062] System failure;

[0063] Triggered a system takeover request;

[0064] The vehicle is about to exceed the system's operational design area;

[0065] When the system detects the following situations, it will issue a takeover request. The takeover request is intended to remind the driver that the situation on the scene requires the driver to take over the steering wheel, and at the same time collect the driver's driving behavior. The collection of the driver's driving behavior is used to determine whether the driver is suitable for rapid and comprehensive takeover of the steering wheel. By collecting the relevant physiological characteristics of the driver and comparing them with the system's built-in model, the driver's current driving status is confirmed, and a trust level value containing the driver's information is generated based on the driving status, providing an operational basis for the subsequent system behavior.

[0066] The following combination Figure 3 , details how the system generates the corresponding trust model based on the trust level value:

[0067] S21. Collect the generated trust level values;

[0068] S22, determining and evaluating the score range of the trust level value, and generating a preliminary model that conforms to the trust level value;

[0069] S23. Confirm and refine the model into a trust model adapted to the current driving state through a mapping function.

[0070] In an embodiment of the present invention, the system obtains the driver's driving information by receiving the aforementioned trust level value; then the system analyzes the trust level value interval, where the system preset interval within which the trust level value falls is analyzed to generate a rough function model; the trust level value is calculated by the function model to generate a corresponding trust model; the mapping function here is g(τ, t), where τ is the trust level value and t is the time process. Based on the mapping function g(τ, t), a weight parameter λd(t) for determining the driver's steering input and the steering input of the autonomous driving system are also generated. The input weight parameter λc(t) is used to generate the weight parameter, which is distributed in real time through the weight distribution strategy. λc(t) represents the specific performance of the steering input of the autonomous driving system in the trust model based on the function: δf=λd(t)δd+λc(t)δc, where δf is the steering angle of the vehicle's front wheels, δc is the steering wheel input of the autonomous driving system, and δd is the steering wheel input of the driver. After the system completes the real-time weight distribution, the steering angle of the vehicle's front wheels is determined by the autonomous driving end and the driving end. The trust model generated above can be divided into three levels: low, medium, and high:

[0071] Figure 4In the illustrated trust model, when the trust level is low, the lateral control weights of the driver and the automated driving system change more slowly. This is because at a low trust level, the system tends to be more conservative and safe. Therefore, changes in control weights are made more cautiously and slowly for both the driver and the automated driving system to avoid potential danger or uncertainty. When drivers lack trust in the automated driving system's performance, they tend to retain more control authority, causing the system to be more cautious in taking lateral control actions to align with the driver's intentions and expectations.

[0072] Figure 5 The trust model shown is at a medium level, where both the driver and the automated driving system's lateral control weights change relatively smoothly. Drivers at this level generally prefer a comfortable driving experience and have a certain degree of confidence in the system's performance. Therefore, the system makes smoother changes in lateral control weights to ensure smooth and stable driving. At this level, drivers are more likely to accept the system making faster lateral control adjustments when necessary to improve driving efficiency and smoothness while maintaining safety. This trade-off results in relatively smooth and effective changes in control weights.

[0073] Figure 6 When the trust model shown is at a high trust level, the lateral control weights of the driver and the autonomous driving system change relatively quickly, and the driver can quickly gain more control of the vehicle. A high trust level means that the driver has great confidence in the functions and performance of the autonomous driving system. Therefore, when the driver decides to take over control, the system will respond quickly and hand over more control to allow the driver to operate the vehicle independently. A high trust level is usually accompanied by the system's efficient real-time response capabilities. The system can quickly identify the driver's intentions and actions, so when the driver begins to take over, the system can immediately adjust the lateral control weight. Systems with a high trust level can often achieve a higher level of driving experience while ensuring safety. This includes the driver's ability to quickly and effectively adjust the lateral control weight to respond to various driving scenarios and changes in road conditions without causing an unstable or uncomfortable driving experience;

[0074] The generated trust models all possess the characteristic of λc(t) + λd(t) = 1. That is, regardless of the direction of the generated trust model, the sum of the driver's steering input weight λd(t) and the autonomous driving system's steering input weight λc(t) always sums to 1. As can be seen from the foregoing, the driver's steering input weight λd(t) and the autonomous driving system's steering input weight λc(t) change in real time, following a time-varying pattern: the driver's control weight gradually approaches 1, while the autonomous driving system's weight gradually approaches 0. This means that the driver will fully assume control of the steering wheel at a predetermined time, while the autonomous driving system will fully release control, completing a flexible takeover. When the autonomous driving system releases control of the steering wheel, the system also releases longitudinal control authority, such as the accelerator and brake. Once the driver confirms that they have taken over the steering wheel, they will also take over longitudinal control authority after the system has released it, completing the takeover of vehicle control.

[0075] like Figure 7 As shown, an embodiment of the present invention further provides a conditional automatic driving takeover device based on human-machine trust, the system comprising:

[0076] The trust level generating module S81 is used to collect the driving behavior and driving environment information of the driver on the driving side, and make judgments based on the driving behavior to generate a trust level value;

[0077] A trust model generation module S82 is used to collect the generated trust level values, determine the score interval of the trust level values ​​and generate a trust model;

[0078] The weight allocation module S83 is used to generate a weight allocation strategy based on the trust model and change the weights in real time according to the weight allocation module;

[0079] The takeover module S84 is used to detect the weight distribution and determine the takeover conditions. If the takeover conditions are met, the lateral control authority is released and the lateral control authority is allocated to the driver. After the lateral authority handover is completed, the longitudinal control authority is released and the longitudinal control authority is allocated to the driver.

[0080] An embodiment of the present invention also provides a conditional autonomous driving takeover system device based on human-machine trust, including a memory, a processor, and a computer program that can be run on the processor. When the processor executes the program, any of the above embodiments can be implemented; the terminal where the processor, memory, and program are deployed here can be a computer, a mobile terminal, or a server. Only the preferred carrier of the loading system is proposed here, and the device is not limited.

[0081] The above are only specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. The embodiments of the present invention and the features therein can be combined with each other unless there is a conflict. Therefore, the scope of protection of the present invention shall be based on the scope of protection of the claims.

Claims

1. A conditional autonomous driving takeover method based on human-machine trust, characterized in that: include: Collect driving environment information and the driver's driving behavior, make judgments based on the driving behavior, and generate a trust level value; Collect the generated trust level values, determine the score interval of the trust level values ​​and generate a trust model; Generate a weight distribution strategy based on the trust model, and adjust the steering wheel input weights on the driver side and the autonomous driving side in real time based on the weight distribution module; Detect the weight distribution and determine the takeover conditions. If the takeover conditions are met, release the lateral control authority and assign the lateral control authority to the driver. After the lateral control authority is transferred, release the longitudinal control authority and assign the longitudinal control authority to the driver. The collecting of generated trust level values, determining the score interval of the trust level values ​​and generating a trust model further includes: collecting the generated confidence level values; determining a score interval of the confidence level value, and generating a preliminary mapping function that conforms to the score interval of the confidence level value; By calculating the mapping function, a trust model adapted to the current driving state is generated; The mapping function used to generate the trust model is: g(τ,t) Among them, τ is the trust level value, t is the time process, Based on the mapping function g(τ,t), a weight parameter λd(t) for determining the driver's steering input and a weight parameter λc(t) for the autonomous driving system's steering input are generated in a score interval corresponding to the trust level value. The weight of the driver's steering input λd(t) and the weight of the autonomous driving system's steering input λc(t) satisfy the following conditions: λc(t) + λd(t)=1 That is, the sum of the weight of the driver's steering input and the weight of the autonomous driving system's steering input always sums to 1.

2. The method for conditional autonomous driving takeover based on human-machine trust according to claim 1, characterized in that: The collecting of driving environment information and the driving behavior of the driver on the driving side, and making judgments based on the driving behavior to generate a trust level value also includes: Collect the driver's driving behavior and driving environment information; By collecting relevant physiological characteristics of the driver's driving behavior and comparing them with the system model; defining the driver's driving state according to the comparison result; A confidence level value is generated based on the driving status.

3. The method for conditional autonomous driving takeover based on human-machine trust according to claim 1, characterized in that: The method of generating a weight distribution strategy based on the trust model and changing the weights in real time based on the weight distribution module also includes: The real-time change trend of the weight is gradually biased towards the driving side, and completely biased towards the driving side at the point preset by the system.

4. The method for conditional autonomous driving takeover based on human-machine trust according to claim 1, characterized in that: The detection weight distribution situation is determined, and the takeover conditions are determined. If the takeover conditions are met, the lateral control authority is released and the lateral control authority is allocated to the driver end. After the lateral authority handover is completed, the longitudinal control authority is released and the longitudinal control authority is allocated to the driver end. The method also includes: δf = λd (t)δd + λc(t)δc δf is the steering angle of the vehicle's front wheels, δc is the steering wheel input of the autonomous driving system, and δd is the steering wheel input of the driver. δf is determined collaboratively by the driver and the autonomous driving system. When the value of λd is 1, the vehicle is in manual driving. When the value of λc is 1, the vehicle is in manual automatic operation, and the lateral control authority is released and assigned to the driver to complete the takeover. At the same time, the longitudinal control authority is also released and assigned to the driver to complete the takeover.

5. Conditional autonomous driving takeover equipment based on human-machine trust, characterized by: The method comprises a memory, a processor and a computer program that can be run on the processor. When the processor executes the program, a conditional autonomous driving vehicle takeover method based on human-machine trust as claimed in any one of claims 1 to 4 is implemented.

Images