An off-road unmanned vehicle trajectory planning method based on ACT-R cognitive decision

Abstract

The present disclosure proposes a kind of field unmanned vehicle trajectory planning method based on ACT-R cognitive decision-making;The method comprises the following steps: global planning based on improved A algorithm, wherein the evaluation function of improved A algorithm introduces logarithmic weighting function to construct environment complexity adaptive heuristic function, and delete the redundant node in the planning path process of improved A algorithm;Local planning based on improved DWA algorithm;Wherein, the evaluation function of improved DWA algorithm is improved DWA that fuses multiple indexes, the indexes of the improved DWA evaluation function include the angle deviation of the direction of the end of the current trajectory and the final target point, the angle deviation of the direction of the end of the current trajectory and the sub-target point, the Euclidean distance from the end of the current trajectory to the nearest obstacle and various weight coefficients;Various weight coefficients are dynamically adjusted based on ACT-R cognitive decision-making.The present application can be beneficial to the problem that unmanned vehicle trajectory planning process is easy to fall into local optimal solution.

Description

Technical Field

[0001] This disclosure relates to the field of unmanned vehicle path planning technology, and in particular to a method for planning the trajectory of unmanned vehicles in the wild based on ACT-R cognitive decision-making. Background Technology

[0002] As a typical representative of intelligent unmanned systems, unmanned vehicles (RVs) are seeing their application areas continuously expand, and are now widely used in natural disaster relief, wilderness exploration, and other outdoor environments. The wilderness environment is a highly complex, unstructured space with irregular terrain, characterized by significant topographic relief, numerous irregular obstacles such as rocks and gullies, and vastly different physical properties of the surface materials. These complex conditions place extremely high demands on the environmental perception, autonomous navigation, and decision-making control capabilities of RVs in the wild, and efficient and safe trajectory planning technology is crucial for their autonomous operation.

[0003] Motion planning typically employs a two-tiered architecture that combines global planning and local planning: global planning is responsible for searching for collision-free spatial paths and providing guidance constraints for local planning; local planning, guided by the global path, responds in real time to dynamic obstacles and generates executable reference trajectories that satisfy kinematic constraints.

[0004] A As a mainstream global path planner, the algorithm has been widely used due to its completeness and search efficiency. However, when dealing with complex field environments, traditional A / B algorithms... The shortcomings of the algorithm are becoming increasingly apparent: its heuristic function has fixed weights, making it prone to getting trapped in local optima in complex field environments; the geometric paths it generates often contain a large number of redundant nodes, resulting in poor path smoothness, which leads to reduced execution efficiency and increased control difficulty.

[0005] In terms of local planning, the Dynamic Window (DWA) method generates executable local reference trajectories by sampling within the velocity space and strictly adhering to kinematic constraints such as the speed and acceleration of the unmanned vehicle in the wild. However, the evaluation function of the traditional DWA algorithm lacks global path guidance, and its fixed weight allocation and poor environmental adaptability make it prone to getting trapped in local optima in complex obstacle groups in the wild.

[0006] How to solve the problem that unmanned vehicles are prone to getting stuck in local optima during trajectory planning in complex field environments is an important problem that urgently needs to be solved in this field. Summary of the Invention

[0007] The purpose of this disclosure is to provide a field unmanned vehicle trajectory planning method based on ACT-R cognitive decision-making to address the shortcomings of existing technologies. It can avoid the problem of UAV trajectory planning getting stuck in local optima in complex field environments.

[0008] This disclosure provides a method for trajectory planning of unmanned vehicles in the wild based on ACT-R cognitive decision-making, which includes the following steps: Based on improved A The algorithm performs global planning, in which the improvement A The algorithm's evaluation function introduces a logarithmic weighted function to construct a heuristic function that adapts to environmental complexity, and for A... The algorithm plans the path and removes redundant nodes. Local planning is performed based on the improved DWA algorithm. The evaluation function of the improved DWA algorithm is an improved DWA that integrates multiple indicators. The indicators of the improved DWA evaluation function include the angle deviation between the current trajectory end direction and the final target point, the angle deviation between the current trajectory end direction and the sub-target point, the Euclidean distance from the trajectory end to the nearest obstacle, and various weight coefficients. The weight coefficients of each item are dynamically adjusted based on ACT-R cognitive decision-making.

[0009] The above-described ACT-R-based cognitive decision-making method for unmanned vehicle trajectory planning in the wild, optionally includes an improved A... The evaluation function of the algorithm is: in, For the evaluation function, For heuristic functions, From the starting point to the current node The actual cost, These are scale adjustment parameters that are dynamically adjusted based on environmental complexity.

[0010] The above-described ACT-R-based cognitive decision-making method for unmanned vehicle trajectory planning in the wild, optionally includes an improved A... The method for deleting redundant nodes during the global planning process of the algorithm is as follows: Get three consecutive nodes on the global path in sequence, and if the three nodes are collinear, delete the node in the middle position.

[0011] In the above-described ACT-R cognitive decision-making-based field unmanned vehicle trajectory planning method, optionally, during the local planning process based on the improved DWA algorithm, an evaluation function of the improved DWA algorithm is constructed based on the constraints of the kinematic model, so as to select the optimal path for execution through the evaluation function of the improved DWA algorithm.

[0012] The above-described ACT-R-based cognitive decision-making method for unmanned vehicle trajectory planning in the wild may optionally include an improved DWA algorithm with the following evaluation function: in, For linear velocity, Angular velocity, For smoothing coefficients, , , , , , These are the weighting coefficients for each item; This represents the angular deviation between the current trajectory's end direction and the final target point. This represents the angular deviation between the current trajectory's end direction and the sub-target point. The distance from the end of the current trajectory to the nearest obstacle is Euclidean distance. This is the straight-line distance from the end of the current trajectory to the endpoint; This is the vertical distance from the end of the current trajectory to the global path.

[0013] The above-described ACT-R cognitive decision-based trajectory planning method for unmanned vehicles in the wild may optionally include dynamically adjusting the weight coefficients of each item based on ACT-R cognitive decision-making. A decision framework integrating global guidance and local obstacle avoidance under cognitive decision-making is pre-constructed; the decision framework dynamically adjusts weights by accumulating historical trajectory information HIST, thereby achieving a dynamic trade-off between global path tasks and local obstacle avoidance constraints. The weighting coefficients are dynamically adjusted based on the assessment weights of trajectory safety margin, path fit, and driving efficiency.

[0014] The above-described ACT-R-based cognitive decision-making method for unmanned vehicle trajectory planning in the wild may optionally include the following adjustment rules for the weight coefficients: When the nearest distance to an obstacle is below the safety threshold, the weighting coefficient is increased. To enhance obstacle avoidance tendency and reduce weighting coefficient To reduce the weight of approaching the endpoint; When the vertical distance of the global path or the deviation of the sub-target orientation angle is too large, the weighting coefficient is increased. and weighting coefficients To guide local trajectories to fit the global path and reduce weighting coefficients To reduce the speed weight; When the safety distance meets the standard but the path smoothness is insufficient, increase the weighting coefficient. And adjust the weighting coefficients and weighting coefficients The proportions are optimized to improve the smoothness of the driving direction and reduce swaying.

[0015] The above-described ACT-R-based cognitive decision-making method for unmanned vehicle trajectory planning in the wild optionally includes dynamically adjusting the weight coefficients based on a safe distance. The weight coefficient adjustment function is as follows: in, This represents the minimum value of the weight; The maximum value of the weight. For safe distance threshold, This is the adjustment coefficient.

[0016] The above-described ACT-R cognitive decision-making-based field unmanned vehicle trajectory planning method includes, optionally, historical trajectory information including action state and corresponding planning state. The action state includes trajectory generation and weight adjustment, and the planning state includes evaluation function weights and trajectory feature values. By using historical trajectory information, it can be determined whether a specific weight adjustment strategy has been used, in order to avoid infinite loops in the decision-making process and provide data support for dynamic adjustments.

[0017] Compared with the prior art, this disclosure has the following beneficial effects: This invention improves A The algorithm performs global planning, transforming the traditional A The algorithm's evaluation function was modified to incorporate a logarithmically weighted heuristic function that adapts to environmental complexity, and the improved A function was removed. Redundant nodes in the algorithm's path planning process are addressed by introducing a logarithmically weighted heuristic function that adapts to environmental complexity. This dynamically reduces the heuristic coefficient in complex regions with dense obstacles to expand the search range, while increasing the coefficient in simpler regions to improve search efficiency. Combined with redundant node removal, this generates a streamlined and smooth global geometric path. Dynamic weight adjustment based on ACT-R cognitive decision-making allows for continuous adjustment of weight coefficients during local planning, preventing the algorithm from getting trapped in local optima. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of this disclosure, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 Here is a flowchart of the steps of the method proposed in this disclosure; Figure 2 A schematic diagram showing three nodes collinear; Figure 3 This is a schematic diagram showing that the three nodes are not collinear; Figure 4 A flowchart illustrating the dynamic adjustment of weights in ACT-R cognitive decision-making. Figure 5 This is a diagram of the kinematic model of the unmanned vehicle proposed in this disclosure. Detailed Implementation

[0020] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0021] In the description of this invention, it should be understood that the terms "longitudinal," "lateral," "thickness," "upper," "lower," "top," "bottom," "inner," "outer," and "radial," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, unless otherwise stated, "a plurality of" means two or more. In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installed," "connected," and "linked" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal communication of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0022] Regarding the problems pointed out in the background technology, the main reasons why traditional trajectory planning methods are prone to getting trapped in local optimization are: First, the heuristic function weights are fixed, which makes them prone to getting trapped in local optima in complex field environments; Second, in terms of local planning, the dynamic window method lacks global path guidance, has fixed weight allocation, poor environmental adaptability, and is very prone to getting trapped in local optima in complex obstacle groups in the field.

[0023] Please see Figure 1 This disclosure provides a method for planning the trajectory of an unmanned vehicle in the wild based on ACT-R cognitive decision-making, which includes the following steps: S1, based on improved A The algorithm performs global planning, in which the improvement A The algorithm's evaluation function introduces a logarithmic weighted function to construct a heuristic function that adapts to environmental complexity, and removes the improved A... Redundant nodes in the algorithm's path planning process. In practical implementation, the operating environment of the unmanned vehicle in the wild is first determined, including the specific locations of obstacles, and the starting point and target point are given. Traditional A The algorithm introduces a heuristic function. Guide the search direction based on the evaluation function Gradually expand the nodes. In this embodiment, improve A... The algorithm mainly improves the evaluation function, specifically targeting the traditional A To address the issue of fixed weights in heuristic functions, this invention improves the evaluation function by introducing a logarithmic weighted function to construct a heuristic function that adapts to environmental complexity. The improved evaluation function is as follows: in, Indicates the distance from the starting point to the current node. The actual cost, Represents a node Heuristic cost estimation to the target node This is a scale adjustment parameter dynamically adjusted based on environmental complexity, used to control the magnitude of heuristic function weight changes; in areas with dense obstacles and complex environments, it is increased... Reducing the heuristic function weights makes the algorithm more reliant on actual path costs, increasing the search range and improving obstacle avoidance capabilities; in open and simple areas, by reducing... This enhances the heuristic function and improves search efficiency. In other words, the scaling parameters, which are dynamically adjusted based on environment complexity, can be customized to meet specific needs.

[0024] In addition, traditional A The algorithm's path planning process suffers from a large number of redundant nodes. This disclosure also proposes a key node selection strategy: sequentially obtain three consecutive nodes on the global path, and delete the node in the middle position when the three nodes are collinear. For example, suppose the three consecutive nodes on the global path are P, M, and Q. If the three points are collinear, such as... Figure 2 As shown, M is a redundant node and is deleted, and the path set is updated accordingly; if the three points are not collinear, as... Figure 3 As shown, it determines whether there is an obstacle between P and Q: if there is an obstacle, M is retained as a key node; if there is no obstacle, M is deleted. This process is repeated until the endpoint node is reached, generating a simplified and smooth global geometric path.

[0025] S2, Local planning is performed based on an improved DWA algorithm. The evaluation function of this improved DWA algorithm is a multi-index fusion improved DWA, whose indices include the angular deviation between the current trajectory's end direction and the final target point, the angular deviation between the current trajectory's end direction and the sub-target point, the Euclidean distance from the trajectory's end to the nearest obstacle, and various weighting coefficients. Specifically, a kinematic model of the autonomous vehicle is established based on geometric relationships. This model ignores the vehicle's mass, inertia, and the complex mechanical interactions between the tires and the ground, thus better describing the autonomous vehicle's position. ,direction linear velocity and angular velocity The dynamic changes. Let... , For driverless cars Given the linear velocity and angular velocity at time intervals, and the infinitesimal displacement within the sampling period Δt, approximating uniform linear motion, the kinematic model expression is given. The principle of the kinematic model can be found in [reference needed]. Figure 5 : DWA comprehensively considers three types of conditions: speed range constraints, acceleration / deceleration constraints, and safe braking distance constraints. It limits the feasible speed to a dynamic window, generates several sets of feasible trajectories within the constraints, scores each trajectory using an evaluation function, and selects the one with the highest score as the optimal path to execute.

[0026] Traditional DWA evaluation function The current DWA evaluation function consists only of three components: heading angle, linear velocity, and closest distance to obstacles. Lacking global path guidance, it is prone to deviation from the global path during long-distance navigation. To address this deficiency, this embodiment constructs an improved DWA evaluation function that integrates multiple indicators: In the formula, This represents the angular deviation between the current trajectory's end direction and the final target point. This refers to the angular deviation between the trajectory's end direction and the sub-target point. The distance from the end of the trajectory to the nearest obstacle is the Euclidean distance. This is the straight-line distance from the end of the trajectory to the endpoint; This is the vertical distance from the end of the trajectory to the global path; For smoothing coefficients, , , , , , These are the weighting coefficients for each item.

[0027] In practical implementation, to effectively transfer global path information to the local DWA algorithm, this embodiment introduces the concept of a dynamic sub-target point (Pref). This dynamic sub-target point is located in the improved A... The generated global path maintains a certain aiming distance from the current position of the autonomous vehicle. In the evaluation function... and By guiding and constraining the local trajectory through this sub-target point, global path information can be integrated with the local path, avoiding path deviation caused by local optimization.

[0028] Specifically, the rules for introducing dynamic sub-target points Pref in this invention are as follows: 1. First, improve A The algorithm calculates the complete global planning path, and the selection range of the dynamic sub-target point Pref is limited to this global path and will not deviate from the global path.

[0029] 2. The dynamic sub-target point Pref and the real-time current position of the unmanned vehicle maintain a preset aiming distance. As the unmanned vehicle continues to move forward, the dynamic sub-target point Pref will dynamically update its position along the global path, always maintaining the aiming distance constraint.

[0030] The main functions of setting dynamic sub-target points (Prefs) are: 1. To play a role in information transmission and avoid losing the global path direction. Dynamic sub-target points (Prefs) are the information bridge between the upper-level global path planning module and the lower-level local DWA dynamic planning module. Relying on dynamic sub-target points (Prefs), improved A... The output global path information is stably transmitted to the local DWA algorithm, overcoming the shortcomings of traditional DWA which only reads the destination information and loses the global path direction; secondly, it constrains the local trajectory, and the Head2(v,) function in the DWA evaluation function... ), Dist3(v, Both metrics are constructed based on dynamic sub-target points (Pref): The dynamic sub-target points (Pref) guide and constrain each local candidate trajectory to be evaluated, ensuring that the local trajectory generation process continuously references the global path. Third, it integrates global and local paths and avoids path deviations; relying on the dynamic sub-target points (Pref) to link Head2(v, ), Dist3(v, The two evaluation indicators achieve deep integration of global path information and local rolling planning path; avoiding the problem of traditional DWA that deviates significantly from the global planning route by only taking the destination as the goal and simply pursuing the optimal local obstacle avoidance.

[0031] S3, dynamically adjust the weight coefficients of each item based on ACT-R cognitive decision-making. In specific steps, this step includes steps S31 and S32.

[0032] Step S31: A decision framework integrating global guidance and local obstacle avoidance under cognitive decision-making is pre-constructed. In this embodiment, the core idea of ​​historical memory-driven decision-making in the ACT-R framework is extracted to construct a decision framework integrating global guidance and local obstacle avoidance under cognitive decision-making. This decision framework dynamically adjusts weights by accumulating historical trajectory information (HIST), achieving a dynamic trade-off between global path tasks and local obstacle avoidance constraints. In this embodiment, historical trajectory information includes action states and corresponding planning states; the action states include trajectory generation and weight adjustment, and the planning states include evaluation function weights and trajectory feature values. Historical trajectory information is used to determine whether a specific weight adjustment strategy has been used, thus avoiding infinite loops during the decision-making process and providing data support for dynamic adjustment.

[0033] ACT-R continuously accumulates historical trajectory information during the autonomous vehicle's operation. This information includes the autonomous vehicle's action and planning states in different wilderness scenarios, which can be represented as: In the formula, For the first Real-time action status, including operation records such as trajectory generation and weight adjustment; The HIST (Hypertext Analyzer) represents the planning state at each given moment, including evaluation function weights and trajectory feature values. It determines whether a specific weight adjustment strategy has been used, preventing infinite loops in the decision-making process and providing data support for dynamic adjustments. The HIST serves as both a memory and training data repository, guiding cognitive decision-making. After each loop, the system evaluates the weight combination based on trajectory safety margin, path fit, and driving efficiency, enabling dynamic learning and optimization of the weight adjustment strategy.

[0034] Step S32 involves evaluating weights based on trajectory safety margin, path fit, and driving efficiency to dynamically adjust the weight coefficients. In practice, the adjustment rules for the weight coefficients include: When the nearest distance to an obstacle is below the safety threshold, the weighting coefficient is increased. To enhance obstacle avoidance tendency and reduce weighting coefficient To reduce the weight of approaching the endpoint; When the vertical distance of the global path or the deviation of the sub-target orientation angle is too large, the weighting coefficient is increased. and weighting coefficients To guide local trajectories to fit the global path and reduce weighting coefficients To reduce the speed weight; When the safety distance meets the standard but the path smoothness is insufficient, increase the weighting coefficient. And adjust the weighting coefficients and weighting coefficients The proportions are optimized to improve the smoothness of the driving direction and reduce swaying.

[0035] When adjusting the weighting coefficients, the weighting coefficients are dynamically adjusted based on the safety distance. The weighting coefficient adjustment function is as follows: in, for The minimum value of the weighting coefficient it represents; for The maximum value of the weighting coefficient referred to. For safe distance threshold, Here, is the adjustment coefficient; where, Placeholder term, referring to , , , , and Any one of them.

[0036] In practice, based on the new weight combination, the DWA algorithm re-evaluates and selects the optimal trajectory to control vehicle execution. The execution result will be used as the new tuple. The data is stored in the HIST, thus completing one full cycle of ACT-R, and the system has implemented the corresponding learning process. The dynamic weight adjustment process for ACT-R cognitive decision-making is as follows: Figure 4 As shown. Specifically, to implement step S1 using the global path planning module, step S2 is implemented using the local trajectory planning module, and step S3 is implemented using the cognitive decision-making module, the coefficients of each processor are dynamically adjusted. In practical applications, the tracking control module can be used to achieve coordinated use between steps S2 and S3.

[0037] It should be noted that in this embodiment, steps S1, S2 and S3 do not necessarily refer to a specific order. For example, steps S2 and S3 can be performed simultaneously or alternately.

[0038] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0039] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for trajectory planning of unmanned vehicles in the wild based on ACT-R cognitive decision-making, characterized in that, Includes the following steps: Based on improved A The algorithm performs global planning, in which the improvement A The algorithm's evaluation function introduces a logarithmic weighted function to construct a heuristic function that adapts to environmental complexity, and removes the improved A... Redundant nodes in the algorithm's path planning process; Local planning is performed based on the improved DWA algorithm. The evaluation function of the improved DWA algorithm is an improved DWA that integrates multiple indicators. The indicators of the improved DWA evaluation function include the angle deviation between the current trajectory end direction and the final target point, the angle deviation between the current trajectory end direction and the sub-target point, the Euclidean distance from the current trajectory end to the nearest obstacle, and various weight coefficients. The weight coefficients of each item are dynamically adjusted based on ACT-R cognitive decision-making.

2. The field unmanned vehicle trajectory planning method based on ACT-R cognitive decision-making as described in claim 1, characterized in that, Based on improved A The evaluation function of the algorithm is: in, For the evaluation function, For heuristic functions, From the starting point to the current node The actual cost, These are scale adjustment parameters that are dynamically adjusted based on environmental complexity.

3. The field unmanned vehicle trajectory planning method based on ACT-R cognitive decision-making as described in claim 1, characterized in that, Based on improved A The method for deleting redundant nodes during the global planning process of the algorithm is as follows: Get three consecutive nodes on the global path in sequence, and if the three nodes are collinear, delete the node in the middle position.

4. The field unmanned vehicle trajectory planning method based on ACT-R cognitive decision-making as described in claim 1, characterized in that, In the process of local planning based on the improved DWA algorithm, an evaluation function for the improved DWA algorithm is constructed based on the constraints of the kinematic model, so as to select the optimal path for execution through the evaluation function of the improved DWA algorithm.

5. The field unmanned vehicle trajectory planning method based on ACT-R cognitive decision-making according to claim 4, characterized in that, The evaluation function of the improved DWA algorithm is: in, Linear velocity, Angular velocity, For smoothing coefficients, , , , , , These are the weighting coefficients for each item; This represents the angular deviation between the current trajectory's end direction and the final target point. This represents the angular deviation between the current trajectory's end direction and the sub-target point. The distance from the end of the current trajectory to the nearest obstacle is Euclidean distance. This is the straight-line distance from the end of the current trajectory to the endpoint; This is the vertical distance from the end of the current trajectory to the global path.

6. The field unmanned vehicle trajectory planning method based on ACT-R cognitive decision-making as described in claim 5, characterized in that, Dynamically adjusting the weight coefficients of each item based on ACT-R cognitive decision-making includes: A decision framework integrating global guidance and local obstacle avoidance under cognitive decision-making is pre-constructed; the decision framework dynamically adjusts weights by accumulating historical trajectory information HIST, thereby achieving a dynamic trade-off between global path tasks and local obstacle avoidance constraints. The weighting coefficients are dynamically adjusted based on the assessment weights of trajectory safety margin, path fit, and driving efficiency.

7. The field unmanned vehicle trajectory planning method based on ACT-R cognitive decision-making as described in claim 6, characterized in that, The adjustment rules for the weighting coefficients include: When the nearest distance to an obstacle is below the safety threshold, the weighting coefficient is increased. To enhance obstacle avoidance tendency and reduce weighting coefficient To reduce the weight of approaching the endpoint; When the vertical distance of the global path or the deviation of the sub-target orientation angle is too large, the weighting coefficient is increased. and weighting coefficients To guide local trajectories to fit the global path and reduce weighting coefficients To reduce the speed weight; When the safety distance meets the standard but the path smoothness is insufficient, increase the weighting coefficient. And adjust the weighting coefficients and weighting coefficients The proportions are optimized to improve the smoothness of the driving direction and reduce swaying.

8. The field unmanned vehicle trajectory planning method based on ACT-R cognitive decision-making according to claim 6, characterized in that, The weighting coefficients are dynamically adjusted based on the safety distance. The weighting coefficient adjustment function is as follows: in, This represents the minimum value of the weight; The maximum value of the weight. For safe distance threshold, This is the adjustment coefficient.

9. The field unmanned vehicle trajectory planning method based on ACT-R cognitive decision-making according to claim 6, characterized in that, Historical trajectory information includes the action status and the corresponding planning status; The action state includes trajectory generation and weight adjustment, and the planning state includes evaluation function weights and trajectory feature values. By using historical trajectory information, it can be determined whether a specific weight adjustment strategy has been used, in order to avoid infinite loops in the decision-making process and provide data support for dynamic adjustments.

Images