Behavior component fusion and stable takeover control method and system based on structural constraint

Abstract

The application discloses a behavior component fusion and stable takeover control method and system based on structure constraint, and relates to the fields of robot control, automation control and artificial intelligence control. The application acquires system state and calculates a risk index, triggers takeover when the risk exceeds a preset threshold and the system state continuously deteriorates, and switches to a fusion control mode. In the takeover state, a control output is generated through weighted combination of at least one pair of opposite behavior components, and the weight is continuously and dynamically adjusted based on the behavior conflict relationship instead of discrete switching. When the system state meets the stability criterion within a continuous number of control cycles, the takeover is automatically exited to restore the original control. The system is connected to the original system as an additional module, and the original controller does not need to be replaced, which can improve the stability of extreme states, avoid collision and maintain task continuity, and is suitable for various unmanned systems such as robots and automatic driving.

Description

Technical Field

[0001] This invention relates to the fields of robot control systems, automated control systems, and artificial intelligence control, and in particular to a control method and system that achieves stable takeover through behavioral component fusion and modal switching when the system is approaching an unstable or risky state. Background Technology

[0002] With the rapid development of Industry 4.0 and artificial intelligence technologies, unmanned systems such as mobile robots, autonomous vehicles, and drones have been widely applied in various fields, including industrial production, logistics and distribution, transportation, and power line inspection. Their operating environments are also gradually extending from structured, closed scenarios to unstructured, open, and complex ones. In these complex scenarios, unmanned systems not only need to efficiently complete pre-set tasks but also must possess the ability to cope with unexpected risks and ensure their own and the surrounding environment's safety. This places unprecedentedly high demands on the stability, robustness, and task continuity of the control system.

[0003] The control systems of existing robots and unmanned systems are mainly divided into two categories: traditional control methods and learning-based control methods. Traditional control methods include PID control, model predictive control, and sliding mode control. These methods rely on accurate system dynamics and environmental models, achieving high precision and stability under normal operating conditions with high model matching. However, when the system has parameter uncertainties, strong external disturbances, or dynamic environmental changes, the performance of traditional control methods deteriorates sharply. For example, PID control is difficult to tune parameters in nonlinear, strongly coupled systems and struggles to adapt to changing conditions; the computational complexity of model predictive control increases exponentially with the increase in the prediction time domain and control dimension, making it difficult to meet real-time requirements in high-speed unmanned systems. Learning-based control methods include reinforcement learning, imitation learning, and deep neural network control. These methods do not require accurate system models and can learn complex control strategies through data-driven approaches, showing significant advantages in handling nonlinear and uncertain problems. However, these methods have inherent drawbacks such as poor generalization ability and insufficient interpretability. Under boundary conditions not covered by the training set, they are prone to unpredictable control behavior, even leading to system malfunction.

[0004] Both of the aforementioned methods struggle to make stable and continuous decisions when faced with conflicting tasks such as approaching a target and avoiding obstacles. Existing technologies often employ hard-switching safety mechanisms based on fixed thresholds to address safety issues. When the distance between the system and an obstacle falls below a preset threshold, the original control signal is directly severed, forcibly switching to pure obstacle avoidance mode or triggering an emergency stop. While this mechanism can avoid collisions to some extent, it can cause sudden deceleration, stopping, or steering of the system. This not only interrupts the execution of the original task, leading to decreased production efficiency or task failure, but may also trigger secondary risks in dynamic environments. For example, a sudden stop of an autonomous vehicle in complex road conditions could cause a rear-end collision, and a sudden stop of an industrial robot on a production line could paralyze the entire production line, resulting in significant economic losses.

[0005] Therefore, there is currently no stable takeover mechanism that can ensure system operation safety while achieving smooth switching of control modes and maintaining continuous task execution. There is an urgent need to develop a new control method to solve the above-mentioned technical problems. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of existing control systems, such as poor stability under complex boundary conditions, unstable decision-making in conflict tasks, and easy interruption of tasks by threshold-based safety mechanisms, and to provide a behavioral component fusion and stable takeover control method and system based on structural constraints.

[0007] To achieve the above objectives, the present invention provides the following technical solution: The behavior component fusion and stable takeover control method based on structural constraints includes the following steps: Step 1: Obtain system status and calculate risk indicators; Step 2: Determine whether to enter the takeover state based on the risk indicators. When the risk indicators exceed the preset threshold and the system status shows a continuous deterioration trend, the takeover control mechanism is triggered, and the system switches from the original control mode to the fusion control mode. Step 3: In the fusion control mode, the control output is generated by weighted combination of multiple behavioral components; Step 4: The behavioral component includes at least one pair of opposing behavioral components; Step 5: The control output adjusts the weights based on the conflict relationship between the opposing behavioral components. The change in weights is driven by the degree of conflict between the behavioral components. The degree of conflict is used as a core variable in the weight function calculation, so that the weights form a continuous evolution process within the control cycle, rather than discrete switching based on a fixed threshold. Step 6: The weights are continuous functions that change dynamically with the system state, rather than being discretely switched; Step 7: When the system state meets the stability criterion for several consecutive control cycles, automatically exit the takeover control and restore the original control mode.

[0008] Furthermore, the control output satisfies the formula: ; in, For the first Each behavioral component For the first The weights corresponding to each behavioral component.

[0009] Furthermore, the opposing behavioral components include a target-approaching component and a risk-avoiding component, and the behavioral components also include a modulating component for suppressing oscillations and smoothing behavior.

[0010] Furthermore, the weights are determined jointly by the degree of risk, the degree of behavioral conflict, and the system stability index.

[0011] Furthermore, the risk indicators include the degree of conflict or structural inconsistency between physical state quantities and behavioral components, specifically the distance to obstacles, the angle between the directions of behavioral components, and the degree of discontinuity in behavioral changes. The behavioral conflict degree is a structural conflict index, which includes at least one or more combinations of behavioral component direction angle, behavioral trend inconsistency, historical change offset of control output, and local stability change rate of the system.

[0012] A structurally constrained behavior component fusion and stability takeover control system includes: The status assessment unit is used to acquire the system status and calculate risk indicators. The takeover decision unit is used to determine whether to enter the takeover state based on risk indicators. When the risk indicators exceed the preset threshold and the system state shows a continuous deterioration trend, takeover control is triggered. When the system state meets the stability criterion within a certain number of consecutive control cycles, takeover control is triggered to exit. The behavior generation unit is used to generate a control output by weighted combination of multiple behavior components in the fusion control mode. The behavior components include at least one pair of opposing behavior components. The control output is weighted based on the conflict relationship between the opposing behavior components. The weights are continuous functions and change dynamically with the system state rather than being discretely switched.

[0013] Furthermore, the system is connected to the original control system as an independent additional control module, and can take over and restore the system without replacing the original controller. It receives the original control output and outputs a corrected control signal. The final control output is a function of the original control output and the system state.

[0014] Furthermore, the system state acquired by the state assessment unit includes system position, system velocity, obstacle position, and target position.

[0015] Furthermore, the approaching target component generated by the behavior generation unit satisfies Stay away from risk factors and satisfy The basic weight calculation satisfies ,in For the target location, This is the current system location. Location of the obstacle. This represents the distance between the system and the obstacle.

[0016] Furthermore, the takeover decision unit has built-in hysteresis switching logic, and there is a difference between the takeover threshold and the recovery threshold, which avoids the system frequently switching control modes in critical states.

[0017] Compared with the prior art, the present invention has the following advantages: 1. Improve system stability in extreme states: When the system is close to instability or task conflict, the control strategy is automatically adjusted by continuous fusion of behavioral components to effectively avoid control oscillations or system divergence. 2. Achieve dynamic obstacle avoidance: By weighted fusion of opposing behavioral components, obstacle avoidance is achieved. Compared with the fixed threshold mechanism, the obstacle avoidance process is smoother and without abrupt movements. 3. Ensure task continuity: Unlike traditional threshold-based safety mechanisms that are prone to task interruption, this invention can automatically and seamlessly switch back to the original control mode after obstacle avoidance is completed and the system returns to stability, and continue to execute the original task. 4. Strong compatibility: This system can be connected to the original control system as an independent add-on module without replacing the original controller. It only needs to receive the original control output and system status signals to achieve stable takeover, resulting in low modification costs. 5. Wide range of applications: It can be widely used in various unmanned systems and automated control systems such as industrial robots, service robots, autonomous vehicles, and drones.

[0018] 6. This invention uses a conflict-driven continuous weight evolution mechanism to ensure that the control output remains continuously differentiable under conflict conditions, thus avoiding the control discontinuity and system oscillation problems caused by threshold switching in traditional control methods. Attached Figure Description

[0019] Figure 1 Here is the overall flowchart of the behavior component fusion and stable takeover control method based on structural constraints; Figure 2 A flowchart of the core mechanism for behavioral component fusion control; Figure 3A comparative diagram illustrating the control of smooth mode switching. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0021] The behavioral component fusion and stable takeover control method based on structural constraints proposed in this invention includes the following steps: Step 1: Obtain system status and calculate risk indicators; Step 2: Determine whether to enter the takeover state based on the risk indicators. When the risk indicators exceed the preset threshold and the system status shows a continuous deterioration trend, the takeover control mechanism is triggered, and the system switches from the original control mode to the fusion control mode. Step 3: In the fusion control mode, the control output is generated by weighted combination of multiple behavioral components; Step 4: The behavioral component includes at least one pair of opposing behavioral components; Step 5: The control output is weighted based on the conflict relationship between the opposing behavioral components; Step 6: The weights are continuous functions that change dynamically with the system state, rather than being discretely switched; Step 7: When the system state meets the stability criterion for several consecutive control cycles, automatically exit the takeover control and restore the original control mode.

[0022] Furthermore, the control output satisfies the formula: ; in, For the first Each behavioral component For the first The weights corresponding to each behavioral component.

[0023] Furthermore, the opposing behavioral components include a target-approaching component and a risk-avoiding component, and the behavioral components also include a modulating component for suppressing oscillations and smoothing behavior.

[0024] Furthermore, the weights are determined jointly by the degree of risk, the degree of behavioral conflict, and the system stability index.

[0025] Furthermore, the risk indicators include the degree of conflict or structural inconsistency between physical state quantities and behavioral components, specifically the distance to obstacles, the angle between the directions of behavioral components, and the degree of discontinuity in behavioral changes.

[0026] This invention also proposes a behavior component fusion and stability takeover control system based on structural constraints, including: The status assessment unit is used to acquire the system status and calculate risk indicators. The takeover decision unit is used to determine whether to enter the takeover state based on risk indicators. When the risk indicators exceed the preset threshold and the system state shows a continuous deterioration trend, takeover control is triggered. When the system state meets the stability criterion within a certain number of consecutive control cycles, takeover control is triggered to exit. The behavior generation unit is used to generate a control output by weighted combination of multiple behavior components in the fusion control mode. The behavior components include at least one pair of opposing behavior components. The control output is weighted based on the conflict relationship between the opposing behavior components. The weights are continuous functions and change dynamically with the system state rather than being discretely switched.

[0027] Furthermore, the system is connected to the original control system as an independent additional control module, and can take over and restore the system without replacing the original controller. It receives the original control output and outputs a corrected control signal. The final control output is a function of the original control output and the system state.

[0028] Furthermore, the system state acquired by the state assessment unit includes system position, system velocity, obstacle position, and target position.

[0029] Furthermore, the approaching target component generated by the behavior generation unit satisfies Stay away from risk factors and satisfy The basic weight calculation satisfies ,in For the target location, This is the current system location. Location of the obstacle. This represents the distance between the system and the obstacle.

[0030] Furthermore, the takeover decision unit has built-in hysteresis switching logic, and there is a difference between the takeover threshold and the recovery threshold, which avoids the system frequently switching control modes in critical states.

[0031] The present invention will be further described in detail below with reference to specific embodiments.

[0032] This embodiment provides a method and system for behavioral component fusion and stable takeover control based on structural constraints. The specific implementation steps are as follows: Step 1: System Status Acquisition and Risk Assessment The state assessment unit acquires the system's position x, velocity v, acceleration a, and the positions of surrounding obstacles in real time at a frequency of 100Hz. Target location Identify physical state quantities and calculate the comprehensive risk index (risk). The risk indicators include not only the distance between the system and obstacles. It also includes indicators of the degree of conflict and structural inconsistency between behavioral components, specifically the angle between the directions of the behavioral components. The degree of discontinuity in behavioral changes ; The behavioral conflict degree is a structural conflict index, which includes at least one or more combinations of behavioral component directional angle, behavioral trend inconsistency, historical change offset of control output, and system local stability change rate. Among them, the angle between the directions of the behavioral components To approach the angle between the target component and the risk component, when When the angle is >120°, it is considered a high-conflict state; the degree of discontinuity in behavioral changes. It is the absolute value of the difference between the control output of the current control cycle and the previous control cycle. >0.5 At that time, it was determined that the system was at risk of oscillation; The comprehensive risk index is calculated using a weighted summation method: ,in =0.5、 =0.3、 =0.2 is a preset weighting coefficient, which can be adjusted according to the security priority of different application scenarios. This represents the system's maximum sensing distance. This is the maximum control output of the system.

[0033] Step 2: Takeover Decision and Smooth Modal Switching The takeover decision-making unit will compare the calculated comprehensive risk indicators with the preset takeover threshold. Compare and monitor the changing trends of the system status; In this embodiment, "the system state shows a continuous deterioration trend" is defined as the increment of the risk indicator being greater than 0 for two consecutive control cycles, and the cumulative increment exceeding 0.1. When both of the above conditions are met simultaneously, the takeover control mechanism is triggered. To avoid control shocks during mode switching, the system adopts a linear weighted transition method. In the first three control cycles before triggering the takeover, the final control output is... , where k increases linearly from 0 to 1, to achieve a smooth transition from the original control mode to the fusion control mode.

[0034] Step 3: Calculation of behavioral components and dynamic weight adjustment The behavior generation unit first calculates each basic behavior component, among which the target-approaching component... And away from risk They are respectively: , ; When multiple obstacles exist, the distance-from-risk component corresponding to each obstacle is calculated separately. Then, the components are weighted and summed based on the distances of each obstacle to the system to obtain the total distance-from-risk component. Simultaneously, the behavior generation unit calculates the weight of each behavior component based on the risk level, the degree of behavior conflict, and the system stability index. In this embodiment, an exponential function is used for continuous adjustment. ; in =2m is the distance scaling factor; To further suppress high-frequency oscillations in the control quantity, an adjustment component is introduced. ,in =0.2 is the smoothing coefficient. The current unsmoothed control output, This is the control output of the previous control cycle.

[0035] Step 4: Generate fusion control output The behavior generation unit weights and combines each behavior component with its corresponding weight, and then adds the adjustment component to generate the final fusion control output: ; This control output can preserve the behavior of approaching the target to the greatest extent while ensuring safety, thus achieving a balance between obstacle avoidance and task execution.

[0036] Step 5: Seamlessly restore control after exiting the takeover. The takeover decision unit continuously monitors the system status. When the system simultaneously meets the following three stability criteria and maintains this state for three consecutive control cycles, it automatically triggers the exit from takeover control: 1. The overall risk index is below the recovery threshold. =0.3; 2. The rate of change of the control quantity is less than / s; 3. The deviation between the actual system trajectory and the original planned trajectory is less than 0.1m. Similar to the takeover process, the exit process also adopts a linear weighted transition method. In the first 3 control cycles before exit, k decreases linearly from 1 to 0, realizing a seamless switch from the fusion control mode to the original control mode, and the system continues to execute the original task.

[0037] In this embodiment, the control system is connected to the original control system as an independent additional control module. Its inputs are the output signal of the original controller and the system status signal, and its output is the corrected final control signal. This connection method does not require large-scale modification of the original control system and can be quickly deployed in various existing unmanned systems and automated equipment.

[0038] For example, when this system is applied to an indoor mobile service robot, if a pedestrian suddenly enters while the robot is delivering food, the system will automatically trigger takeover control. By merging the behavior components of approaching the food delivery target and moving away from the pedestrian, the robot will smoothly bypass the pedestrian with a smooth trajectory. The entire process is seamless and without any collisions. After bypassing the pedestrian, the system will automatically return to the original control mode and continue to complete the food delivery task.

[0039] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A behavioral component fusion and stable takeover control method based on structural constraints, characterized in that, Includes the following steps: Step 1: Obtain system status and calculate risk indicators; Step 2: Determine whether to enter the takeover state based on the risk indicators. When the risk indicators exceed the preset threshold and the system status shows a continuous deterioration trend, the takeover control mechanism is triggered, and the system switches from the original control mode to the fusion control mode. Step 3: In the fusion control mode, the control output is generated by weighted combination of multiple behavioral components; Step 4: The behavioral component includes at least one pair of opposing behavioral components; Step 5: The control output adjusts the weights based on the conflict relationship between the opposing behavioral components. The change in weights is driven by the degree of conflict between the behavioral components. The degree of conflict is used as a core variable in the weight function calculation, so that the weights form a continuous evolution process within the control cycle, rather than discrete switching based on a fixed threshold. Step 6: The weights are continuous functions that change dynamically with the system state, rather than being discretely switched; Step 7: When the system state meets the stability criterion for several consecutive control cycles, automatically exit the takeover control and restore the original control mode.

2. The behavioral component fusion and stable takeover control method based on structural constraints according to claim 1, characterized in that, The control output satisfies the formula: ； in, For the first Each behavioral component For the first The weights corresponding to each behavioral component.

3. The behavioral component fusion and stable takeover control method based on structural constraints according to claim 1, characterized in that, The opposing behavioral components include a target-approaching component and a risk-avoiding component, and the behavioral components also include a modulating component for suppressing oscillations and smoothing behavior.

4. The behavioral component fusion and stable takeover control method based on structural constraints according to claim 1, characterized in that, The weights are determined by a combination of risk level, behavioral conflict level, and system stability index.

5. The behavioral component fusion and stable takeover control method based on structural constraints according to claim 1, characterized in that, The risk indicators include the degree of conflict or structural inconsistency between physical state quantities and behavioral components, specifically the distance to obstacles, the angle between the directions of behavioral components, and the degree of discontinuity in behavioral changes.

6. A structural constraint-based behavioral component fusion and stable takeover control system implementing the method of any one of claims 1 to 5, characterized in that, include: The status assessment unit is used to acquire the system status and calculate risk indicators. The takeover decision unit is used to determine whether to enter the takeover state based on risk indicators. When the risk indicators exceed the preset threshold and the system state shows a continuous deterioration trend, takeover control is triggered. When the system state meets the stability criterion within a certain number of consecutive control cycles, takeover control is triggered to exit. The behavior generation unit is used to generate a control output by weighted combination of multiple behavior components in the fusion control mode. The behavior components include at least one pair of opposing behavior components. The control output is weighted based on the conflict relationship between the opposing behavior components. The weights are continuous functions and change dynamically with the system state rather than being discretely switched.

7. The behavior component fusion and stable takeover control system based on structural constraints according to claim 6, characterized in that, The system is connected to the original control system as an independent additional control module, and takes over and restores the system without replacing the original controller. It receives the original control output and outputs a corrected control signal. The final control output is a function of the original control output and the system state.

8. The behavior component fusion and stable takeover control system based on structural constraints according to claim 6, characterized in that, The system state acquired by the state assessment unit includes system position, system velocity, obstacle position, and target position.

9. The behavior component fusion and stable takeover control system based on structural constraints according to claim 6, characterized in that, The target-approaching component generated by the behavior generation unit satisfies Stay away from risk factors and satisfy The basic weight calculation satisfies ,in For the target location, This is the current system location. Location of the obstacle. This represents the distance between the system and the obstacle.

10. The behavior component fusion and stable takeover control system based on structural constraints according to claim 6, characterized in that, The takeover decision unit has built-in hysteresis switching logic, and there is a difference between the takeover threshold and the recovery threshold, which avoids the system frequently switching control modes in critical states.

Images