A behavior tree-based unmanned vehicle action modeling and control method

Abstract

The present invention discloses a behavior tree-based unmanned vehicle action modeling and control method. The method constructs an unmanned vehicle action model according to a visual behavior tree. When the unmanned vehicle action model is used for simulation, when a target enters the sensor detection range, a target detection node transmits brief intelligence. A target identification node identifies the target model and payload to obtain intelligence information. A target threat analysis node analyzes the threat level of the target based on the identification results of the target identification node and provides treatment recommendations. A target tracking node is used to implement mobile tracking or non-mobile tracking based on environmental conditions, concealment capabilities, and target detection capabilities. When mobile tracking is implemented, the target tracking node controls the speed and direction of the unmanned vehicle to achieve target tracking. A cluster collaborative control node is used to achieve collaborative control between the individual unmanned vehicle entities in the unmanned vehicle cluster. The present invention implements the application of behavior trees in an unmanned vehicle simulation system.

Description

Technical Field

[0001] The present invention relates to the field of unmanned vehicle technology, and in particular to an unmanned vehicle action modeling and control method based on a behavior tree. Background Art

[0002] How to generate realistic and credible behaviors is the core issue of CGF (Computer Generated Force) research. Currently, the main behavior modeling technologies include: behavior modeling methods based on finite state machines (FSM), CGF behavior modeling methods based on agents and multi-agents, etc.

[0003] Finite state machines (FSMs) are the most widely used behavioral modeling approach in simulation. However, FSMs limit behavioral autonomy. While they are suitable for describing and modeling low-level physical behaviors, such as movement and weapon firing, they are more difficult to model high-level intelligent behaviors or complex behaviors involving parallel operations.

[0004] Agent-based and multi-agent CGF behavior modeling approaches include SOAR, an agent modeling environment based on symbolic representation and rule-based reasoning developed by the University of Michigan. SOAR is used in multiple CGF systems. However, the limitations of agent-based CGF behavior modeling approaches include a lack of a unified understanding of the concept and definition of agents, and a lack of a clear theoretical framework.

[0005] Behavioral modeling has always been a difficult problem for simulation systems. The use of visual behavioral modeling technology can lower the programming threshold for modelers, give full play to the ability of operators to input and edit behavioral models, and allow trainees to write and visually debug operational behavior rules themselves, which can accelerate the model development process and make operational rules more applicable.

[0006] Behavior trees are an improvement on finite state machines. Due to their modularity, reusability, and extensibility, they are widely used in complex control systems for NPCs (Non-Player-Controlled Characters), drones, and robots in games. Behavior trees can combine simple primitive behaviors into complex and robust operating procedures.

[0007] Behavior trees offer advantages such as modularity, reusability, and good readability. The modularity of behavior trees supports model expansion and makes the modeling process more flexible. The reusability of behavior trees reduces the workload of model development. Each behavior subtree can run independently and can be reused as a subtree of other behavior trees. These modularity and reusability enable them to support planning problems with multiple nodes and large-scale states. At the same time, due to the various logical structures they provide, they still maintain good readability.

[0008] The modeling elements of behavior trees include nodes, return states, and tree traversal.

[0009] (1) Nodes. Behavior tree nodes include root nodes, action node sets, condition node sets, and logic node sets. The root node sends an enable signal (Tick) to its child nodes at a certain frequency, which is passed to the action node or condition node layer by layer through the logic node, and receives the return result of the node. The action node (Action) changes the system state. The condition node (Condition) queries the current state. The logic node controls the execution logic of the entire behavior tree and is responsible for passing the enable signal (Tick) downward and reporting the execution status upward. Commonly used logic nodes include decorator nodes (Decorator), sequence nodes (Sequence), selector nodes (Selector), and parallel nodes (Parallel). The description of each node is as follows:

[0010] (a) Action node: A node used to complete an action or task. Generally, it changes the system state and is usually a leaf node in the behavior tree.

[0011] (b) Condition node: checks whether a given condition is met and returns Success if it is met, otherwise returns Failed;

[0012] (c) Modified nodes: used to specify additional conditions for the execution of the node, such as time interval, number of executions, and frequency;

[0013] (d) Sequence Node: For vertical behavior trees, the execution is from left to right; for horizontal behavior trees, the execution is from top to bottom. For horizontal behavior trees, the execution is from top to bottom. If Failed is returned midway, the sequence node is terminated.

[0014] (e) Selecting a node: For vertical behavior trees, select a child node whose execution condition is true from left to right, and for horizontal behavior trees, select a child node from top to bottom. Upon encountering the first child node that succeeds (Success) or is currently executing (Running), the result of the current child node is immediately returned, and the selection node ends. If all child nodes return Failed, the current execution node also fails. Commonly used are improved selection nodes with priority or probability values, which prioritize child nodes with higher priority or execute child nodes or subtrees based on probability sampling.

[0015] (f) Parallel node: executes all child nodes or subtrees in parallel.

[0016] Behavior tree nodes are simple in design and easy to improve. With the widespread application of behavior trees, many improved node types have emerged, such as sequence nodes and selection nodes with time slices, and nodes with inverters.

[0017] (2) Return result status. The node execution will get the return result, and the return result status includes Success, Failed, Running and Error.

[0018] (a) Success state: The condition of the condition node is met, the behavior of the action node is successfully executed, or the logic node meets the logic of successful execution, and the result is Success;

[0019] (b) Failed state: If the condition of the condition node is not met, the behavior of the action node is not successfully executed, or the logic node does not meet the logic for successful execution, Failed is returned;

[0020] (c) Running state: Generally used for action nodes, indicating that the behavior is in progress but has not yet been completed, and returns to Running;

[0021] (d) Error state: An error occurs during node execution and Error is returned.

[0022] (3) Tree Traversal. In the basic implementation of a behavior tree, the system starts at the root of the tree each frame, checking whether each node is activated from top to bottom and from left to right, until it reaches the currently activated node and refreshes the behavior tree. Commonly used are improved priority or weighted sequential nodes, which execute more important or influential child nodes first.

[0023] The behavior tree logic structure is as follows Figure 1 shown.

[0024] Behavior trees (BT) are a behavioral description technology developed in recent years in the field of game artificial intelligence (Game AI). Compared to FSMs, BT has gained increasing market share in the commercial gaming market due to its advantages such as high modularity, two-step control, and hierarchical description structure. However, there have been no reports of its application in autonomous vehicle simulation systems. Summary of the Invention

[0025] In view of the shortcomings of the existing technology, the present invention aims to provide an unmanned vehicle action modeling and control method based on behavior tree.

[0026] In order to achieve the above object, the present invention adopts the following technical solutions:

[0027] A behavior tree-based unmanned vehicle action modeling and control method includes the following steps:

[0028] S1. Constructing a UAV action model according to a visual behavior tree; the UAV action model includes a target discovery node, a target identification node, a target threat analysis node, a target tracking node, and a cluster collaborative control node;

[0029] S2. When simulating the unmanned vehicle action model established in step S1, when the target enters the sensor detection range, the target detection node transmits brief intelligence; the target identification node identifies the target model and payload and obtains intelligence information; the target threat analysis node analyzes the threat level of the target based on the identification result of the target identification node and gives processing suggestions; the target tracking node is used to achieve mobile tracking or non-mobile tracking based on environmental conditions, concealment capabilities, and target detection capabilities. When mobile tracking is implemented, the target tracking node controls the moving speed and direction of the unmanned vehicle to achieve target tracking; the cluster collaborative control node is used to achieve collaborative control between each unmanned vehicle entity in the unmanned vehicle cluster.

[0030] Furthermore, the threat level analysis result is that the target poses no threat, the target poses a moderate threat, or the target poses a severe threat.

[0031] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program implements the above method when executed by a processor.

[0032] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program implements the above method when executed by a processor.

[0033] The beneficial effect of the present invention is that the present invention realizes the application of behavior trees in unmanned vehicle simulation systems, and gives full play to the advantages of behavior trees in drone action modeling and control. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 This is a diagram of the behavior tree logic structure;

[0035] Figure 2 Execute the schematic diagram for the selected node;

[0036] Figure 3 Execute the schematic diagram for the sequence node;

[0037] Figure 4 Execute the schematic diagram for parallel nodes;

[0038] Figure 5 Execute the schematic for the decoration node;

[0039] Figure 6 This is a detection behavior rule view of the unmanned vehicle action model established according to the visual behavior tree in an embodiment of the present invention. DETAILED DESCRIPTION

[0040] The present invention will be further described below in conjunction with the accompanying drawings. It should be noted that this embodiment is based on the technical solution and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to this embodiment.

[0041] This embodiment provides a behavior tree-based unmanned vehicle behavior modeling and control method, including the following steps:

[0042] S1. Establish the autonomous vehicle action model according to the visual behavior tree;

[0043] The visual behavior tree is a directed tree consisting of nodes and directed edges. The execution of a behavior tree in a simulation cycle is the traversal of control flow from the root node to the leaf nodes. The traversal of control flow from a parent node to a child node is called a tick. After a child node receives and executes the control flow, it must report the execution result to its parent node. The execution results include success, failure, and running. The parent node determines the direction of the control flow by examining the execution result.

[0044] Except for the root node, each node has only one parent node. Behavior trees have four major types of nodes as shown below:

[0045] CompositeNode;

[0046] DecoratorNode;

[0047] ConditionNode;

[0048] Behavior action node (ActionNode).

[0049] 1) Composite nodes are control nodes that can serve as parent nodes and have multiple child nodes. They can be divided into three types according to their composite properties:

[0050] SelectionNode;

[0051] SequenceNode;

[0052] Parallel Node.

[0053] Among them, the execution logic of the selection node is to successfully run one of the multiple child nodes. Its traversal strategy is to traverse until it encounters the first child node that can run successfully. When it receives the successful execution result returned by the child node, it will immediately return this status to the parent node; when the traversed child node returns a failed execution result, the selection node will continue to traverse the next child node; when the traversed child node returns a running execution result, the behavior tree will record the running child node and start traversing directly from the child node in the next simulation cycle. The selection node is logically equivalent to a logical OR operation. The diagram of the selection node execution is as follows Figure 2 shown.

[0054] The serial node is a control node used to execute child nodes sequentially. Its traversal strategy is to traverse until it encounters the first child node that fails to run. When it receives the failed execution result returned by the child node, it will immediately return this status to the parent node; when the traversed child node returns a successful execution result, the serial node will continue to traverse the next child node; when the traversed child node returns a running execution result, the behavior tree will record the running child node and start traversing directly from this child node in the next simulation cycle. The serial node is logically equivalent to the logical AND operation. The serial node execution diagram is as follows Figure 3 shown.

[0055] The traversal strategy of parallel nodes is different from that of selection nodes and control nodes. Parallel nodes always tick all their child nodes at any time in a simulation cycle. When any child node returns a running execution result, the parallel node will return the running status to the parent node after traversing all child nodes. When no child node returns a running execution result and the number of child nodes that return a successful execution result exceeds a given value N, the parallel node returns a successful execution result to the parent node, otherwise it returns a failed execution result. The parallel node execution diagram is shown in the figure below. Figure 4 shown.

[0056] 2) Decorator nodes are used to change the control node return value strategy, control the number of control node traversals, and set execution conditions. After the child node is executed, it performs additional processing on the execution result returned by the child node and then returns it to its parent node. A decorator node can be a parent node, but can only have one child node. The diagram of the decorator node execution is as follows: Figure 5 shown.

[0057] 3) The function of the conditional node is very clear, that is, it returns True when the predefined conditions of the node are met.

[0058] 4) Action nodes are leaf nodes in the behavior tree that actually execute tasks and implement the autonomous vehicle's behavioral capabilities. After each action node receives control flow, it handles the autonomous vehicle's actions within that simulation cycle, including sensing environmental information, calculating action operations, determining the behavior's execution status, and returning the result. When autonomous vehicle actions need to be executed step by step between nodes, technologies such as blackboard can be introduced for simple data exchange.

[0059] In this embodiment, the unmanned vehicle action model includes a target discovery node, a target identification node, a target threat analysis node, a target tracking node, and a cluster collaborative control node;

[0060] S2. When simulating the unmanned vehicle action model established in step S1, when the target enters the sensor detection range, the target node is found to transmit brief intelligence; the target identification node uses different sensor detection methods, such as visible light, infrared, thermal imaging, etc., to identify the target model and payload and other intelligence information; the target threat analysis node analyzes the threat level of the target based on the identification result of the target identification node, and the analysis result is that the target is no threat, the target is a general threat, or the target is a serious threat. For the analysis result of the serious threat to the target, the target threat analysis node can give suggestions for self-processing (processing according to the platform's payload) or only tracking (not having self-processing capabilities) according to needs; the target tracking node is used to achieve mobile tracking or non-mobile tracking (i.e., perspective tracking) according to environmental conditions, concealment capabilities, and target detection capabilities. When implementing mobile tracking, the target tracking node controls the moving speed and direction of the unmanned vehicle to achieve target tracking; the cluster collaborative control node is used to achieve collaborative control between each unmanned vehicle entity in the unmanned vehicle cluster.

[0061] Those skilled in the art can make various corresponding changes and modifications based on the above technical solutions and concepts, and all of these changes and modifications should be included in the scope of protection of the claims of the present invention.

Claims

1. A behavior tree-based unmanned vehicle action modeling and control method, characterized in that: The steps include: S1. Constructing a UAV action model according to a visual behavior tree; the UAV action model includes a target discovery node, a target identification node, a target threat analysis node, a target tracking node, and a cluster collaborative control node; S2. When simulating the unmanned vehicle action model established in step S1, when the target enters the sensor detection range, the target node is found to transmit brief intelligence; The target recognition node identifies the target model and payload and obtains intelligence information; The target threat analysis node analyzes the threat level of the target based on the identification results of the target identification node and provides treatment suggestions; The target tracking node is used to implement mobile tracking or non-mobile tracking based on environmental conditions, concealment capabilities, and target detection capabilities. When implementing mobile tracking, the target tracking node controls the speed and direction of the unmanned vehicle to achieve target tracking; The cluster collaborative control node is used to achieve collaborative control between each unmanned vehicle entity in the unmanned vehicle cluster.

2. The method according to claim 1, characterized in that The threat level analysis results are: target no threat, target general threat, or target severe threat.

3. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 2 is implemented.

4. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 2 is implemented.

Images