An adaptive control barrier function obstacle avoidance method based on dynamic elliptical non-safe set

Abstract

The application discloses an adaptive control obstacle avoidance function avoidance method based on a dynamic elliptical non-safe set, and relates to the field of intelligent vehicle motion control.The method comprises the following steps: S1, collecting information in a perception layer: acquiring state information of a vehicle, state information of an obstacle and environment information; S2, determining a safety parameter in a risk assessment layer: calculating a comprehensive risk score and matching a target safety level; S3, generating an expected control instruction by an upper controller: generating an expected control instruction according to a driving target of the vehicle, which comprises an expected longitudinal acceleration and an expected front wheel steering angle; S4, constructing a dynamic elliptical non-safe set and a safety distance control obstacle function CBF, determining a focus of the dynamic elliptical non-safe set and determining core parameters of the dynamic elliptical non-safe set; S5, solving a final control instruction meeting a safety constraint; and S6, performing an obstacle avoidance action, which comprises receiving, by an execution layer, driving the power system and the steering system of the vehicle to perform acceleration / deceleration and steering operation.

Description

Technical Field

[0001] This application relates to the field of intelligent vehicle motion control, and in particular to an adaptive control obstacle avoidance method based on dynamic elliptic insecure sets, an adaptive control obstacle avoidance system based on dynamic elliptic insecure sets, electronic devices, storage media, and autonomous vehicles. Background Technology

[0002] With the rapid development of autonomous driving technology, ensuring the physical safety of vehicles in increasingly complex urban dynamic environments has become a key bottleneck and research focus for autonomous driving systems. Although existing autonomous driving systems employ advanced perception and decision-making modules, both hierarchical rule-based planning systems and end-to-end deep learning methods still face significant challenges in providing verifiable safety guarantees. Hierarchical systems may struggle to maintain sufficient safety margins across all dynamic interactions due to the incompleteness of predefined rules or the decoupling between upper-level planning and lower-level control. On the other hand, while end-to-end learning methods have shown potential, their "black box" nature makes it difficult to provide formal safety proofs. Their performance is highly dependent on the coverage of training data, and when faced with critical or adversarial scenarios not included in the training set, they may generate unpredictable or even dangerous control commands, leading to collision risks.

[0003] Research in the field of obstacle avoidance control for autonomous driving aims to achieve a balance between efficiency and stability. Current mainstream research methods can be broadly categorized into two types: The first type is machine learning-based methods. These methods typically model autonomous driving and obstacle avoidance tasks as Markov Decision Processes (MDPs) and utilize reinforcement learning to find the optimal control strategy. For example, Kendall et al. pioneered the application of the DRL framework to end-to-end autonomous driving; Yurtsever et al. proposed a hybrid deep reinforcement learning framework; and Muzahid et al. introduced a DRL-based driving strategy to reduce collision risk in high-density traffic flows. The second type is optimization-based techniques. These methods formalize the obstacle avoidance problem as an optimization problem with explicit safety constraints, aiming to compute control inputs that rigorously guarantee safety by minimizing control costs or other performance metrics. Among these, the obstacle function (CBF) has attracted considerable attention as a powerful controller synthesis tool. The core advantage of CBF lies in its ability to provide formal safety guarantees (usually manifested as forward invariance of the safety set), that is, by making minimal necessary interventions when the system approaches the unsafe state boundary, it mathematically proves that the system will remain within a predefined safe region, thus preserving the original performance objectives as much as possible when safety is not threatened. Researchers have successfully applied CBF to the field of autonomous driving and verified its effectiveness, such as for maintaining a safe distance in adaptive cruise control (ACC), combining it with model predictive control (MPC) to enable safe overtaking in racing cars, and using speed obstacles to build a new type of CBF for dynamic obstacle avoidance.

[0004] Therefore, absolute safety is difficult to guarantee in scenarios such as interactive lane changing in traffic flow or negotiated passage at unsignalized intersections. Traditional path and trajectory planning algorithms often fail to meet the requirements of real-time safety control due to solution delays in rapidly changing environments or high-dimensional state spaces. Although some studies have achieved longitudinal safety maintenance, they have failed to provide a comprehensive safety strategy for both lateral and longitudinal directions. Combining the control obstacle function with model predictive control (MPC) can achieve multi-step safety prediction, but the inherent high computational complexity of MPC, coupled with CBF constraints, poses a severe challenge to onboard computing platforms. The method of constructing a new CBF using speed obstacles not only increases the computational burden by optimizing speed and steering control independently, but also results in a lack of clear arbitration mechanism between speed-priority and steering-priority obstacle avoidance strategies due to the complete decoupling of the two. Moreover, when the control obstacle function is constructed based on relative distance, its first-order time derivative is only a function of relative velocity and does not directly include acceleration and angular velocity as control inputs. This structure causes the terms related to control inputs in the standard CBF constraints to always be zero (i.e., L_gh ≡ 0), making the constraint ineffective for control synthesis. Furthermore, the key parameters of traditional control obstacle functions (such as safety margin) are usually static and cannot be dynamically adjusted according to real-time changing scenarios, which weakens the safety robustness of the control system under complex operating conditions. Summary of the Invention

[0005] The purpose of this invention is to provide an adaptive control obstacle avoidance method based on dynamic elliptic insecure sets, an adaptive control obstacle avoidance system based on dynamic elliptic insecure sets, an electronic device, a storage medium, and an autonomous vehicle, thereby solving at least one of a number of technical problems.

[0006] 1. When facing complex dynamic scenarios outside the training distribution or adversarial situations, the limited generalization ability and robustness of end-to-end solutions make it difficult to ensure the continuous safety of control commands, which may lead to physical collisions. In particular, absolute safety cannot be guaranteed in scenarios such as interactive lane changing in traffic flow or negotiated passage at unsignalized intersections.

[0007] 2. Traditional path and trajectory planning algorithms often fail to meet the requirements of real-time safety control due to solution delays in rapidly changing environments or high-dimensional state spaces.

[0008] 3. While combining obstacle control functions with model predictive control (MPC) can achieve multi-step safety prediction, the inherent high computational complexity of MPC, coupled with CBF constraints, poses a severe challenge to onboard computing platforms. The method of constructing a novel CBF using speed obstacles not only increases the computational burden by optimizing speed and steering control independently, but also results in a lack of clear arbitration mechanism between speed-priority and steering-priority obstacle avoidance strategies due to their complete decoupling. Furthermore, when the obstacle control function is constructed based on relative distance, its first-order time derivative is only a function of relative velocity and does not directly include acceleration and angular velocity as control inputs. This structure causes the terms related to control inputs in the standard CBF constraints to always be zero (i.e., L_gh ≡ 0), rendering the constraint ineffective for control synthesis.

[0009] 4. The key parameters of traditional control obstacle functions (such as safety margin) are usually static and cannot be dynamically adjusted according to real-time changing scenarios, which weakens the safety robustness of the control system under complex operating conditions.

[0010] This invention provides the following solution:

[0011] According to a first aspect of the present invention, an adaptive control obstacle avoidance method based on a dynamic elliptic insecure set is provided, applied to the motion control of an intelligent vehicle to achieve safe obstacle avoidance control in dynamic environments, comprising:

[0012] S1. Sensing layer collects information: acquires the vehicle's status information, obstacle status information, and environmental information;

[0013] S2. Risk assessment layer determines safety parameters: calculates comprehensive risk score. ,according to Match the target security level;

[0014] S3. Upper-level controller generates desired control commands: Generates desired control commands based on the vehicle's driving target. , Including expected longitudinal acceleration and expected front wheel steering angle ;

[0015] S4. Construct the dynamic elliptical unsafe set and the safety distance control barrier function CBF, determine the focus of the dynamic elliptical unsafe set and the core parameters of the dynamic elliptical unsafe set;

[0016] S5. Solve for the final control command that satisfies the safety constraints. ;

[0017] S6. Execute obstacle avoidance actions, including the execution layer receiving the final control command. It drives the vehicle's power system and steering system to perform acceleration / deceleration and steering operations, achieving collision-free obstacle avoidance.

[0018] Furthermore, the information collected by S1 and the perception layer includes:

[0019] Vehicle status information includes the vehicle's Cartesian coordinates. longitudinal speed of the vehicle Self-driving heading angle ;

[0020] Obstacle status information includes the obstacle's Cartesian coordinates. longitudinal velocity of the obstacle Obstacle heading angle longitudinal acceleration of obstacles Obstacle yaw rate and the confidence level of the obstacle state.

[0021] Furthermore, the S2 risk assessment layer determines the following safety parameters:

[0022] S21. Calculate the comprehensive risk score. ,in:

[0023] This is a high-speed risk factor, with a default value of 0, when the vehicle's longitudinal speed... Exceeding the preset high-speed threshold hour, ;

[0024] This represents an adverse environmental factor, with a default value of 0. It is triggered when environmental conditions that significantly negatively impact driving safety are detected. ;

[0025] This is the obstacle factor, with a default value of 0. It is used when the obstacle state confidence level is less than the preset confidence threshold. hour, ;

[0026] S22, according to Matching target security levels include: preset There are several security levels, and each security level corresponds to a set of security parameters: and ;

[0027] And security level number The larger, the corresponding Monotonically increasing, Monotonically decreasing, based on Determine the current target security level and its corresponding ;

[0028] in, For safety margin; The coefficients are K-type functions; .

[0029] Furthermore, S4, constructing the dynamic elliptic insecure set and the safe distance control barrier function CBF includes:

[0030] S41. The foci for determining the dynamic elliptic unsafe set include:

[0031] First Focus : Get the current position of the obstacle, including ;

[0032] Second Focus : For dynamic backtracking points, along the current speed of the vehicle Translation in the opposite direction The distance corresponding to the time is obtained, and the translation formula is: , ,in, To predict the time domain, and ;

[0033] S42. The core parameters for determining dynamic elliptic unsafe sets include:

[0034] Ellipse center :Pick and The midpoint, including, , ;

[0035] Elliptical focal length: and The straight-line distance between them;

[0036] The length of the major axis of an ellipse is equal to the sum of the focal length and twice the radius of the ellipse's envelope circle.

[0037] S43. Construct the safe distance CBF function :

[0038] ;

[0039] in, The radius of the envelope circle and the center of the ellipse are given by the following: The sum of distances from the vehicle direction to the boundary of the ellipse.

[0040] Furthermore, S5, solving for the final control instructions that satisfy the safety constraints includes:

[0041] S51. Establish the system kinematic model: Use a nonlinear single-vehicle model, the expression is:

[0042] ;in Let be the system state vector.

[0043] Let be the vehicle's state subvector.

[0044] This is the obstacle state sub-vector;

[0045] Where x=[ , ;

[0046] It is the state vector of the obstacle;

[0047] ,

[0048] ;

[0049] , ,

[0050] This refers to the wheelbase of the vehicle. It is a 4x2 zero matrix;

[0051] S52. Construct a control optimization model:

[0052] Objective function: ;in, The control commands to be solved include the actual longitudinal acceleration. and actual front wheel steering angle ; For weighted norm; This is a pre-defined diagonal weighted matrix; Output to the upper-level controller;

[0053] Constraints: ,in, Preset to positive numbers; for Regarding system drift items Li's derivative; for Regarding the system control influence matrix Li's derivative;

[0054] S53. Solving the control optimization model: Obtain the final control command that satisfies the constraints through quadratic programming. .

[0055] Furthermore, including:

[0056] In step S41, the time domain is predicted. According to step S21 Dynamic adjustment:

[0057] when This includes the vehicle's longitudinal speed not exceeding When no environmental conditions that significantly negatively impact driving safety are detected and the obstacle state confidence level is not less than K, The values ​​are 0.1s and 0.3s;

[0058] when Including the longitudinal speed of the vehicle exceeding When at least two of the following conditions are met: environmental conditions that significantly negatively impact driving safety are detected, and the confidence level of the obstacle state is less than K, The values ​​are 0.5s and 0.8s.

[0059] Furthermore, including:

[0060] Step S4 executes steps S41-S43 for each obstacle separately to generate... CBF function with a safety distance ;

[0061] In step S52, the constraints of the control optimization model are updated to... The objective function remains unchanged, and the final control command satisfying all constraints is obtained by quadratic programming. .

[0062] Furthermore, including:

[0063] In step S3, the upper-level controller is a hierarchical rule-based planning controller or an end-to-end deep learning controller.

[0064] When the upper-level controller outputs The constraints in step S52 are not satisfied. At that time, through secondary planning Make corrections to obtain the final control commands. .

[0065] According to a second aspect of the present invention, an adaptive control obstacle avoidance system based on a dynamic elliptic insecure set is provided, comprising executing an adaptive control obstacle avoidance method based on a dynamic elliptic insecure set, including:

[0066] The layers consist of a perception layer, a risk assessment layer, a control barrier function layer, an execution layer, and an upper-level controller.

[0067] The perception layer is used to acquire the vehicle's status information, obstacle status information, and environmental information.

[0068] Risk assessment layer, used to determine safety parameters;

[0069] The upper-level controller is used to generate the desired control commands;

[0070] The obstacle control function layer is used to receive and fuse safety parameters and desired control commands, optimize and correct them in real time, and output a set of acceleration and front wheel steering angle commands to ensure safety.

[0071] The execution layer is used to drive the vehicle's powertrain and steering system to perform acceleration / deceleration and steering operations, thereby achieving collision-free obstacle avoidance.

[0072] According to a third aspect of the present invention, an electronic device is provided, comprising: a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus;

[0073] The memory stores a computer program that, when executed by the processor, causes the processor to perform the steps of an obstacle avoidance method based on an adaptive control obstacle function of a dynamic elliptic insecure set.

[0074] According to a fourth aspect of the present invention, a computer-readable storage medium is provided storing a computer program executable by an electronic device, which, when run on the electronic device, causes the electronic device to perform the steps of an obstacle avoidance method based on an adaptive control obstacle function of a dynamic elliptic insecure set.

[0075] According to a fifth aspect of the present invention, an autonomous driving vehicle is provided, comprising:

[0076] An electronic device for implementing the steps of an adaptive control obstacle avoidance method based on a dynamic elliptic insecure set;

[0077] The processor runs a program, and when the program runs, it executes the steps of an obstacle avoidance method based on an adaptive control obstacle function using a dynamic elliptic insecure set, based on data output from the electronic device.

[0078] A storage medium for storing a program that, when running, executes steps of an obstacle avoidance method based on an adaptive control function of a dynamic elliptic insecure set in response to data output from an electronic device.

[0079] The above solution achieves the following beneficial technical effects:

[0080] 1. This fundamentally solves the "relative order" failure problem when standard distance CBF is applied to a second-order vehicle dynamic system. Existing technologies directly use the relative distance h as the obstacle function, and its first derivative does not include acceleration and angular velocity inputs, resulting in L_gh always being zero and the constraint being ineffective.

[0081] 1. This application introduces a prediction time domain t IThis design defines h as the distance from the current state to the future dynamic unsafe set, cleverly incorporating the current speed and heading into the expression of h. This design ensures that its time derivative explicitly includes the control inputs a and δ, making L_gh non-zero. This allows the standard CBF framework to directly and effectively constrain the vehicle's lateral and longitudinal control, guaranteeing the safety of control commands. For example, it fundamentally solves the "relative order" failure problem when standard distance CBF is applied to a vehicle's second-order dynamic system. For instance, existing technologies directly use the relative distance h as the obstacle function, whose first derivative does not include acceleration and angular velocity inputs, causing L_gh to always be zero, rendering the constraint ineffective.

[0082] 2. This application achieves effective safety constraints while balancing computational efficiency and control performance. Compared to the MPC method, which can also handle high-order systems but requires enormous computation, the CBF constraints in this application can typically be transformed into a more computationally efficient quadratic programming (QP) problem, ensuring safety. Compared to the method combining velocity obstacle (VO) and CBF, this application unifies the lateral and longitudinal safety constraints within a single obstacle function, avoiding control decoupling and complex priority arbitration problems, and generating a more coordinated and smoother integrated obstacle avoidance trajectory.

[0083] 3. This application introduces a risk assessment mechanism and an adjustable parameter t. I This endows the safety strategy with unprecedented adaptability and flexibility. Traditional safety strategies are often static, while the risk assessment layer of this application can dynamically adjust the safety margin according to the real-time environment and vehicle status. Meanwhile, parameter t... I The system is provided with a clear adjustment knob that allows it to switch between two obstacle avoidance styles: "responsive" and "smooth look-ahead," which greatly enhances the adaptability and overall performance of autonomous vehicles in complex dynamic scenarios. Attached Figure Description

[0084] Figure 1 This is a flowchart of an adaptive control obstacle avoidance method based on a dynamic elliptic insecure set, provided by one or more embodiments of the present invention.

[0085] Figure 2 This is a structural diagram of an adaptive control obstacle avoidance system based on a dynamic elliptic insecure set, provided by one or more embodiments of the present invention.

[0086] Figure 3 This is a schematic diagram of an adaptive control obstacle avoidance architecture based on a dynamic elliptic insecure set, provided by a specific embodiment of the present invention.

[0087] Figure 4 This is a schematic diagram of the construction of a safe distance CBF constraint provided in a specific embodiment of the present invention.

[0088] Figure 5 This is a schematic diagram of a simulation of CBF constraint modification of the upper-level controller hazard control input provided in a specific embodiment of the present invention.

[0089] Figure 6 This is a block diagram of an electronic device structure for an adaptive control obstacle avoidance method based on dynamic elliptic insecure sets, provided by one or more embodiments of the present invention. Detailed Implementation

[0090] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. 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.

[0091] Figure 1 This is a flowchart of an adaptive control obstacle avoidance method based on a dynamic elliptic insecure set, provided by one or more embodiments of the present invention.

[0092] like Figure 1 The method shown is an adaptive control obstacle avoidance method based on dynamic elliptic insecure sets, applied to the motion control of autonomous vehicles to achieve safe obstacle avoidance control in dynamic environments, including:

[0093] S1. Sensing layer collects information: acquires the vehicle's status information, obstacle status information, and environmental information;

[0094] S2. Risk assessment layer determines safety parameters: calculates comprehensive risk score. ,according to Match the target security level;

[0095] S3. Upper-level controller generates desired control commands: Generates desired control commands based on the vehicle's driving target. , Including expected longitudinal acceleration and expected front wheel steering angle ;

[0096] S4. Construct the dynamic elliptical unsafe set and the safety distance control barrier function CBF, determine the focus of the dynamic elliptical unsafe set and the core parameters of the dynamic elliptical unsafe set;

[0097] S5. Solve for the final control command that satisfies the safety constraints. ;

[0098] S6. Execute obstacle avoidance actions, including the execution layer receiving the final control command. It drives the vehicle's power system and steering system to perform acceleration / deceleration and steering operations, achieving collision-free obstacle avoidance.

[0099] In this embodiment, S1, the information collected by the sensing layer includes:

[0100] Vehicle status information includes the vehicle's Cartesian coordinates. longitudinal speed of the vehicle Self-driving heading angle ;

[0101] Obstacle status information includes the obstacle's Cartesian coordinates. longitudinal velocity of the obstacle Obstacle heading angle longitudinal acceleration of obstacles Obstacle yaw rate and the confidence level of the obstacle state.

[0102] In this embodiment, S2, the risk assessment layer determines the safety parameters including:

[0103] S21. Calculate the comprehensive risk score. ,in:

[0104] This is a high-speed risk factor, with a default value of 0, when the vehicle's longitudinal speed... Exceeding the preset high-speed threshold hour, ;

[0105] This represents an adverse environmental factor, with a default value of 0. It is triggered when environmental conditions that significantly negatively impact driving safety are detected. ;

[0106] This is the obstacle factor, with a default value of 0. It is used when the obstacle state confidence level is less than the preset confidence threshold. hour, ;

[0107] S22, according to Matching target security levels include: preset There are several security levels, and each security level corresponds to a set of security parameters: and ;

[0108] And security level number The larger, the corresponding Monotonically increasing, Monotonically decreasing, based on Determine the current target security level and its corresponding ;

[0109] in, For safety margin; The coefficients are K-type functions; .

[0110] In this embodiment, S4, constructing the dynamic elliptic insecure set and the safe distance control barrier function CBF includes:

[0111] S41. The foci for determining the dynamic elliptic unsafe set include:

[0112] First Focus : Get the current position of the obstacle, including ;

[0113] Second Focus : For dynamic backtracking points, along the current speed of the vehicle Translation in the opposite direction The distance corresponding to the time is obtained, and the translation formula is: , ,in, To predict the time domain, and ;

[0114] S42. The core parameters for determining dynamic elliptic unsafe sets include:

[0115] Ellipse center :Pick and The midpoint, including, , ;

[0116] Elliptical focal length: and The straight-line distance between them;

[0117] The length of the major axis of an ellipse is equal to the sum of the focal length and twice the radius of the ellipse's envelope circle.

[0118] S43. Construct the safe distance CBF function :

[0119] ;

[0120] in, The radius of the envelope circle and the center of the ellipse are given by the following: The sum of distances from the vehicle direction to the boundary of the ellipse.

[0121] In this embodiment, S5, solving for the final control command that satisfies the safety constraints includes:

[0122] S51. Establish the system kinematic model: Use a nonlinear single-vehicle model, the expression is:

[0123] ;in Let be the system state vector.

[0124] Let be the vehicle's state subvector.

[0125] This is the obstacle state sub-vector;

[0126] Where x=[ , ;

[0127] It is the state vector of the obstacle;

[0128] ,

[0129] ;

[0130] , ,

[0131] This refers to the wheelbase of the vehicle. It is a 4x2 zero matrix;

[0132] S52. Construct a control optimization model:

[0133] Objective function: ;in, The control commands to be solved include the actual longitudinal acceleration. and actual front wheel steering angle ; For weighted norm; This is a pre-defined diagonal weighted matrix; Output to the upper-level controller;

[0134] Constraints: ,in, Preset to positive numbers; for Regarding system drift items Li's derivative; for Regarding the system control influence matrix Li's derivative;

[0135] S53. Solving the control optimization model: Obtain the final control command that satisfies the constraints through quadratic programming. .

[0136] In this embodiment, it includes:

[0137] In step S41, the time domain is predicted. According to step S21 Dynamic adjustment:

[0138] when This includes the vehicle's longitudinal speed not exceeding When no environmental conditions that significantly negatively impact driving safety are detected and the obstacle state confidence level is not less than K, The values ​​are 0.1s and 0.3s;

[0139] when Including the longitudinal speed of the vehicle exceeding When at least two of the following conditions are met: environmental conditions that significantly negatively impact driving safety are detected, and the confidence level of the obstacle state is less than K, The values ​​are 0.5s and 0.8s.

[0140] In this embodiment, it includes:

[0141] Step S4 executes steps S41-S43 for each obstacle separately to generate... CBF function with a safety distance ;

[0142] In step S52, the constraints of the control optimization model are updated to... The objective function remains unchanged, and the final control command satisfying all constraints is obtained by quadratic programming. .

[0143] In this embodiment, it includes:

[0144] In step S3, the upper-level controller is a hierarchical rule-based planning controller or an end-to-end deep learning controller.

[0145] When the upper-level controller outputs The constraints in step S52 are not satisfied. At that time, through secondary planning Make corrections to obtain the final control commands. .

[0146] Figure 2 This is a structural diagram of an adaptive control obstacle avoidance system based on a dynamic elliptic insecure set, provided by one or more embodiments of the present invention.

[0147] like Figure 2 The obstacle avoidance system based on the adaptive control function of dynamic elliptic insecure sets, as shown, executes the obstacle avoidance method based on the adaptive control function of dynamic elliptic insecure sets, including:

[0148] The layers consist of a perception layer, a risk assessment layer, a control barrier function layer, an execution layer, and an upper-level controller.

[0149] The perception layer is used to acquire the vehicle's status information, obstacle status information, and environmental information.

[0150] Risk assessment layer, used to determine safety parameters;

[0151] The upper-level controller is used to generate the desired control commands;

[0152] The obstacle control function layer is used to receive and fuse safety parameters and desired control commands, optimize and correct them in real time, and output a set of acceleration and front wheel steering angle commands to ensure safety.

[0153] The execution layer is used to drive the vehicle's powertrain and steering system to perform acceleration / deceleration and steering operations, thereby achieving collision-free obstacle avoidance.

[0154] It is worth noting that although this system / device only discloses the aforementioned perception layer, risk assessment layer, control barrier function layer, execution layer, and upper-level controller, it does not mean that this system / device is limited to the aforementioned basic functional modules. Rather, what this invention intends to express is that, based on the aforementioned basic functional modules, those skilled in the art can arbitrarily add one or more functional modules in combination with existing technology to form an infinite number of embodiments or technical solutions. In other words, this system is open rather than closed. The fact that this embodiment only discloses a few basic functional modules should not lead one to believe that the scope of protection of the claims of this invention is limited to the aforementioned disclosed basic functional modules.

[0155] In one specific embodiment, an adaptive obstacle avoidance architecture based on dynamic elliptic insecure sets is disclosed:

[0156] like Figure 3 As shown, the system architecture begins at the perception layer, where the acquired obstacle information is synchronously transmitted to the upper-level controller and risk setting layer. The risk setting layer calculates the safety margin and key K-type function coefficients based on this information, while the upper-level controller generates the desired acceleration and front wheel steering angle commands according to its own strategy. The core obstacle control function layer receives and integrates these two inputs, performs real-time optimization and correction, and outputs a set of safe acceleration and front wheel steering angle commands. This set of safety-verified final commands is then handed over to the execution layer, driving the system to complete the specific physical operations.

[0157] The risk assessment layer incorporates a hierarchical decision-making model. Its core function is to calculate a comprehensive risk score S_risk in real time and select from four preset security levels based on this score. Each security level corresponds to a specific set of security parameters (security margin dei, K-type function coefficient γi), where i∈{1,2,3,4}. As the index i increases, the security margin monotonically increases and the coefficient monotonically decreases, representing a more conservative and robust security strategy. The comprehensive risk score S_risk is obtained by summing the quantitative scores of three core dimensions—self-state (A), environmental state (B), and key objective (C). S_risk = Score(A) + Score(B) + Score(C)

[0158] Score(A): High-speed risk factor, with a default value of 0. When the vehicle speed exceeds a preset high-speed threshold v_max, the risks of high kinetic energy and long braking distance brought about by high speed are considered to be activated, i.e., Score(A)=1.

[0159] Score(B): Environmental adverse factor, with a default value of 0. When the system detects any environmental conditions that have a significant negative impact on driving safety, the environmental adverse risk is considered to be activated, i.e., Score(B)=1.

[0160] Score(C): Obstacle factor, default value is 0. When the confidence of the perceived obstacle state information is less than the threshold K, the critical obstacle risk is activated, that is, Score(C)=1.

[0161] In another specific embodiment, an affine system is disclosed for accurately modeling the motion of a vehicle and dynamic obstacles in a world coordinate system, such as using a nonlinear vehicle model to describe each entity.

[0162] The system is described as follows:

[0163] ;

[0164] in It is the state vector of the vehicle;

[0165] It is the state vector of the obstacle;

[0166] Control input In the state vector, and Cartesian coordinates representing the center of gravity of the vehicle and the obstacle, respectively;

[0167] φvobs describes their longitudinal velocity; φego and φobs represent their direction. In the control input, Describes the longitudinal acceleration of the host vehicle, Describes the front wheel steering angle. For a static obstacle, all motion-related parameters vobs, φobs, and βobs are all zero. The specific expressions of the drift vector components fego1(x) and fobs1(x) and the control influence matrix Gego(x) are as follows:

[0168] ; ; ;

[0169] where L describes the wheelbase, and and correspond to the predicted longitudinal acceleration and predicted yaw rate (angular velocity) of the obstacle, respectively.

[0170] In another specific embodiment, a safety distance CBF constraint construction is disclosed, which constructs a dynamic elliptical non-safe area that can reflect the motion trend of the host vehicle.

[0171]

[0172]

[0173]

[0174]

[0175]

[0176] This embodiment designs a new safety distance CBF and constructs a dynamic elliptical non-safe area that can reflect the motion trend of the host vehicle. The two foci of the ellipse are determined by the current position p1(x p1 , y p1 ) of the obstacle and a dynamic backtracking point p2(x p2 , y p2 ). The backtracking point p2 is obtained by driving p1 in the opposite direction of the current speed of the host vehicle for t1 (0s < t1 < 1s). As Figure 4 and shown in the above formula, the center p3 of the ellipse takes the midpoint of p1 and p2, its focal length is the distance between p1 and p2, and the length of the major axis is equal to the sum of the focal length and twice the radius of the envelope circle.

[0177] where is the radius of the host vehicle circle, the distance from the center of the ellipse along the vehicle direction to the boundary, de is the safety margin, is the coordinate of the host vehicle in the Cartesian coordinate system, is the speed and heading angle of the host vehicle.

[0178] Introduce speed and heading into the obstacle function, and The time derivative of the control barrier function will explicitly include the control inputs α and δ. This makes the predictive barrier function have an effective first-order relative degree for collision avoidance-related system dynamics, thus allowing the CBF condition to directly constrain the control input.

[0179] like Figure 5 As shown in the simulation, the upper-level controller uses a proportional controller without obstacle avoidance capabilities. Under unconstrained conditions, its control commands will cause the vehicle to collide directly with obstacles. After correction using a self-designed CBF safety constraint, the vehicle can still achieve collision-free safe driving, but the system's obstacle avoidance behavior varies depending on the selected t1. As shown in the figure, smaller t1 values ​​(such as 0.1s, 0.3s) trigger more reactive avoidance actions, characterized by later obstacle avoidance initiation, more abrupt maneuvers, and a smaller minimum safe distance from obstacles. Conversely, larger t1 values ​​(such as 0.5s, 0.8s) prompt the system to adopt a more proactive and smoother obstacle avoidance strategy, enabling earlier intervention and maintaining a greater safety margin throughout the process.

[0180] Within the safety constraint framework, the road environment exhibits complexity due to its multi-element composition. A typical characteristic is the presence of multiple obstacles, which possess both static and dynamic attributes. Therefore, the controller optimization model constructed for such complex scenarios has several constraints in its corresponding controller optimization equations:

[0181] It is a control command that establishes the constraint. These are the control commands from the original controller. Final control command, Represents constraint code, It is a constraint quantity.

[0182] Figure 6 This is a block diagram of an electronic device structure for an adaptive control obstacle avoidance method based on dynamic elliptic insecure sets, provided by one or more embodiments of the present invention.

[0183] like Figure 6 As shown, this application provides an electronic device, including: a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus;

[0184] The memory stores a computer program that, when executed by a processor, causes the processor to perform steps of an obstacle avoidance method based on an adaptive control function of a dynamic elliptic insecure set.

[0185] This application also provides a computer-readable storage medium storing a computer program executable by an electronic device, which, when run on the electronic device, causes the electronic device to perform steps of an obstacle avoidance method based on an adaptive control obstacle function of a dynamic elliptic insecure set.

[0186] This application also provides an autonomous driving vehicle, including:

[0187] An electronic device for implementing the steps of an adaptive control obstacle avoidance method based on a dynamic elliptic insecure set;

[0188] The processor runs a program, and when the program runs, it executes the steps of an obstacle avoidance method based on an adaptive control obstacle function using a dynamic elliptic insecure set, based on data output from the electronic device.

[0189] A storage medium for storing a program that, when running, executes steps of an obstacle avoidance method based on an adaptive control function of a dynamic elliptic insecure set in response to data output from an electronic device.

[0190] The communication bus mentioned in the above electronic devices can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not indicate that there is only one bus or one type of bus.

[0191] The electronic device comprises a hardware layer, an operating system layer running on top of the hardware layer, and an application layer running on the operating system. The hardware layer includes hardware such as a central processing unit (CPU), a memory management unit (MMU), and memory. The operating system can be any one or more computer operating systems that control the electronic device through processes, such as Linux, Unix, Android, iOS, or Windows. Furthermore, in this embodiment of the invention, the electronic device can be a smartphone, tablet computer, or other handheld device, or a desktop computer, portable computer, or other electronic device; there is no particular limitation in this embodiment.

[0192] In this embodiment of the invention, the executing entity for electronic device control can be an electronic device itself, or a functional module within an electronic device capable of calling and executing a program. The electronic device can obtain the firmware corresponding to the storage medium. This firmware is provided by the supplier, and different storage media may have the same or different firmware; no limitation is made here. After obtaining the firmware corresponding to the storage medium, the electronic device can write this firmware into the storage medium; specifically, it burns the firmware corresponding to the storage medium into the storage medium. The process of burning the firmware into the storage medium can be implemented using existing technology, and will not be elaborated upon in this embodiment of the invention.

[0193] Electronic devices can also obtain reset commands corresponding to the storage media. The reset commands corresponding to the storage media are provided by the supplier. The reset commands corresponding to different storage media can be the same or different, and no restrictions are imposed here.

[0194] At this time, the storage medium of the electronic device is a storage medium on which the corresponding firmware has been written. The electronic device can respond to the reset command corresponding to the storage medium on which the corresponding firmware has been written, thereby resetting the storage medium on which the corresponding firmware has been written according to the reset command. The process of resetting the storage medium according to the reset command can be implemented by existing technology and will not be described in detail in this embodiment of the invention.

[0195] For ease of description, the above devices are described separately by function as various units and modules. Of course, in implementing this application, the functions of each unit and module can be implemented in one or more software and / or hardware.

[0196] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the meaning consistent with their meaning in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless specifically defined.

[0197] For the sake of simplicity, the method embodiments are described as a series of actions. However, those skilled in the art should understand that the embodiments of the present invention are not limited to the described order of actions, because according to the embodiments of the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions involved are not necessarily essential to the embodiments of the present invention.

[0198] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments of this application.

[0199] 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 or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An adaptive control obstacle avoidance method based on dynamic elliptic insecure sets, applied to the motion control of autonomous vehicles, to achieve safe obstacle avoidance control in dynamic environments, characterized in that... include: S1. Sensing layer collects information: acquires the vehicle's status information, obstacle status information, and environmental information; Vehicle status information includes the vehicle's Cartesian coordinates. longitudinal speed of the vehicle Self-driving heading angle ; S2. Risk assessment layer determines safety parameters: calculates comprehensive risk score. ,according to Match the target security level; Each security level corresponds to a set of security parameters: and ; in, For safety margin; The coefficients are K-type functions; S3. Upper-level controller generates desired control commands: Generates desired control commands based on the vehicle's driving target. The Including expected longitudinal acceleration and expected front wheel steering angle ; S4. Construct the dynamic elliptical unsafe set and the safety distance control barrier function CBF, determine the focus of the dynamic elliptical unsafe set and the core parameters of the dynamic elliptical unsafe set; Constructing the dynamic elliptic insecure set and the safe distance control barrier function (CBF) includes: S41. The foci for determining the dynamic elliptic unsafe set include: First Focus : Get the current position of the obstacle, including ; Second Focus : For dynamic backtracking points, along the current speed of the vehicle Translation in the opposite direction The distance corresponding to the time is obtained, and the translation formula is: , ,in, To predict the time domain, and ; S42. The core parameters for determining dynamic elliptic unsafe sets include: Ellipse center :Pick and The midpoint, including, , ; Elliptical focal length: and The straight-line distance between them; The length of the major axis of the ellipse is equal to the sum of the focal length and twice the radius of the self-contained envelope circle; S43. Construct the safe distance CBF function : ； in, The radius of the envelope circle and the center of the ellipse are given by the following: The sum of distances from the vehicle direction to the boundary of the ellipse; S5. Solve for the final control command that satisfies the safety constraints. ; S6. Execute obstacle avoidance actions, including the execution layer receiving the final control command. It drives the vehicle's power system and steering system to perform acceleration / deceleration and steering operations, achieving collision-free obstacle avoidance.

2. The obstacle avoidance method based on dynamic elliptic insecure sets and adaptive control obstacle function according to claim 1, characterized in that, The information collected by S1 and the sensing layer includes: The obstacle state information includes the obstacle's Cartesian coordinates. longitudinal velocity of the obstacle Obstacle heading angle longitudinal acceleration of obstacles Obstacle yaw rate and the confidence level of the obstacle state.

3. The obstacle avoidance method based on the adaptive control function of dynamic elliptic insecure sets according to claim 2, characterized in that, The safety parameters determined by the S2 risk assessment layer include: S21. Calculate the comprehensive risk score. ,in: This is a high-speed risk factor, with a default value of 0, when the vehicle's longitudinal speed... Exceeding the preset high-speed threshold hour, ; This represents an adverse environmental factor, with a default value of 0. It is triggered when environmental conditions that significantly negatively impact driving safety are detected. ; This is the obstacle factor, with a default value of 0. It is used when the obstacle state confidence level is less than the preset confidence threshold. hour, ; S22, according to Matching target security levels include: preset There are several security levels, and each security level corresponds to a set of security parameters: and ; And security level number The larger, the corresponding Monotonically increasing, Monotonically decreasing, based on Determine the current target security level and its corresponding ; in, .

4. The obstacle avoidance method based on dynamic elliptic insecure sets and adaptive control obstacle function according to claim 3, characterized in that, S5, solving for the final control command that satisfies the safety constraints, includes: S51. Establish the system kinematic model: Use a nonlinear single-vehicle model, the expression is: ;in Let be the system state vector. Let be the vehicle's state subvector. This is the obstacle state sub-vector; Where x=[ , ; It is the state vector of the obstacle; ， ； ， ， This refers to the wheelbase of the vehicle. It is a 4x2 zero matrix; S52. Construct a control optimization model: Objective function: ;in, The control commands to be solved include the actual longitudinal acceleration. and actual front wheel steering angle ; For weighted norm; This is a pre-defined diagonal weighted matrix; Output to the upper-level controller; Constraints: ,in, Preset to positive numbers; for Regarding system drift items Li's derivative; for Regarding the system control influence matrix Li's derivative; S53. Solving the control optimization model: Obtain the final control command that satisfies the constraints through quadratic programming. .

5. The obstacle avoidance method based on dynamic elliptic insecure sets and adaptive control obstacle function according to claim 4, characterized in that, include: In step S41, the time domain is predicted. According to step S21 Dynamic adjustment: when This includes the vehicle's longitudinal speed not exceeding When no environmental conditions that significantly negatively impact driving safety are detected and the obstacle state confidence level is not less than K, The values ​​are 0.1s and 0.3s; when Including the longitudinal speed of the vehicle exceeding When at least two of the following conditions are met: environmental conditions that significantly negatively impact driving safety are detected, and the confidence level of the obstacle state is less than K, The values ​​are 0.5s and 0.8s.

6. The obstacle avoidance method based on dynamic elliptic insecure sets and adaptive control obstacle function according to claim 5, characterized in that, include: Step S4 executes steps S41-S43 for each obstacle separately to generate... CBF function with a safety distance ; In step S52, the constraints of the control optimization model are updated to... The objective function remains unchanged, and the final control command satisfying all constraints is obtained by quadratic programming. .

7. The obstacle avoidance method based on dynamic elliptic insecure sets and adaptive control obstacle function according to claim 6, characterized in that, include: In step S3, the upper-level controller is a hierarchical rule-based planning controller or an end-to-end deep learning controller. When the upper-level controller outputs The constraints in step S52 are not satisfied. At that time, through secondary planning Make corrections to obtain the final control commands. .

8. An adaptive control obstacle avoidance system based on a dynamic elliptic insecure set, executing the adaptive control obstacle avoidance method based on a dynamic elliptic insecure set as described in any one of claims 1 to 7, characterized in that, include: The layers consist of a perception layer, a risk assessment layer, a control barrier function layer, an execution layer, and an upper-level controller. The perception layer is used to acquire the vehicle's status information, obstacle status information, and environmental information. Risk assessment layer, used to determine safety parameters; The upper-level controller is used to generate the desired control commands; The obstacle control function layer is used to receive and fuse safety parameters and desired control commands, optimize and correct them in real time, and output a set of acceleration and front wheel steering angle commands to ensure safety. The execution layer is used to drive the vehicle's powertrain and steering system to perform acceleration / deceleration and steering operations, thereby achieving collision-free obstacle avoidance.

9. An electronic device, characterized in that, include: The processor, communication interface, memory, and communication bus are connected, with the processor, communication interface, and memory communicating with each other via the communication bus. The memory stores a computer program that, when executed by a processor, causes the processor to perform the steps of the obstacle avoidance method based on the adaptive control obstacle function of a dynamic elliptic insecure set as described in any one of claims 1 to 7.

Images