A multi-UAV collaborative early warning and safety control method based on risk triggering

Abstract

The present invention discloses a multi-UAV collaborative early warning and safety control method based on risk triggering, including: constructing a safety loss model between UAVs and external static and / or dynamic obstacles, and between multiple UAVs; quantitatively evaluating the collision risk of the multi-UAV system based on the safety loss model and conditional value at risk (CVaR); designing a collaborative early warning mechanism, so that even if a certain UAV itself does not perceive an obstacle, it can still obtain obstacle risk information through collaborative behavior; designing a distributed safety controller based on risk triggering, which is composed of a basic formation control item, an inter-machine collision avoidance control item, and an obstacle avoidance control item. This method can break through the limitations of the individual perception ability of UAVs and significantly improve the safety of the multi-UAV system facing dynamic obstacles; at the same time, it can reduce the obstacle avoidance maneuvering distance, save energy and effort, and reduce conservatism. This method is suitable for the collaborative safety flight control of a multi-UAV system in a complex and dense obstacle environment.

Description

Technical Field

[0001] The present invention belongs to the technical field of UAV cluster decision-making and control, and specifically relates to a multi-UAV collaborative early warning and safety control method based on risk triggering, which is suitable for collaborative safe flight control of multi-UAV systems in complex and dense obstacle environments. Background Art

[0002] Thanks to the rapid development of automated control and artificial intelligence technology, multi-UAV systems have been widely used in various fields, such as agricultural applications, monitoring and detection, search and rescue missions, etc. However, the problem of safe flight control of multi-UAV systems in complex and dense obstacle environments remains a core technical challenge. Many studies on UAVs focus on improving their flight safety. In the actual application of multi-UAV systems, various risk factors such as multi-source composite interference, actuator / sensor failure, malicious attacks, and external obstacles pose a serious threat to their safety, which will lead to a significant reduction in system performance and even cause catastrophic safety accidents. In response to these risk factors, many experts and scholars have proposed effective countermeasures. For example, the composite hierarchical anti-interference control theory can specifically separate, estimate, learn and compensate for different interferences in the control system, thereby promoting the design of a series of low-conservative fine controllers with high robustness and high-precision control goals. In addition, there are also cutting-edge technologies such as robust control and fault-tolerant control, which have good effects on improving the safety of multi-UAV systems under different harsh conditions.

[0003] Autonomous obstacle avoidance for external obstacles and collision avoidance between multiple drones are one of the most important safety issues for multi-drone systems. Although many different obstacle avoidance technologies have been proposed and applied, such as potential field method, linear / nonlinear model predictive control, etc., the existing technologies usually rely heavily on the perception distance and range of individual drones, and can only achieve rapid avoidance of obstacles in front of the drone. In addition, the existing obstacle avoidance control strategies are usually based on deterministic safety distance maintenance or related constraints. However, the drone system's own state estimation and perception information of external obstacles are affected by various noises and interferences, and have strong uncertainty. Although the method based on deterministic settings can effectively improve the safety of drone systems, it is also highly conservative. In order to scientifically balance the safety and conservatism of multi-drone systems during flight, uncertainty / probabilistic information should be fully considered. Since chance constraints have the intuitive and practical role of limiting the probability of unsafe events, they are widely used to describe uncertainty in safety issues. However, chance constraints are non-convex and difficult to solve in practical applications. Therefore, scholars have developed conditional risk value based on chance constraints, that is, the conditional expectation of loss distribution within the worst-case quantile. Conditional Value at Risk has been widely used in safety analysis of robot motion planning and control under risk conditions. In addition, safety analysis based on conditional value at risk can distinguish the worst-case tail events, that is, rare but extremely unsafe events are taken into account, which will further improve the safety of multi-UAV systems.

[0004] In summary, although there are many methods available to achieve safe obstacle avoidance control of multi-UAV systems, they are limited by the individual perception capabilities of UAVs and cannot achieve safe avoidance of dynamic obstacles outside the individual perception range. In addition, the existing methods do not fully consider uncertainty, and the obstacle avoidance maneuver distance is large and conservative. Summary of the invention

[0005] In order to overcome the defects of the existing methods and address the problem of safe flight of multiple UAV systems in complex and dense obstacle environments, the present invention provides a multi-UAV collaborative early warning and safety control method based on risk triggering. The present invention can ensure the flight safety and survivability of UAV formations in dense static and / or dynamic obstacle environments.

[0006] In order to achieve the above object, the present invention adopts the following technical scheme:

[0007] A multi-UAV collaborative early warning and safety control method based on risk triggering includes the following steps:

[0008] In the first step, considering the uncertainty of system state estimation and obstacle perception, a safety loss model between UAVs and external static and / or dynamic obstacles, as well as between multiple UAVs, is constructed.

[0009] The second step is to quantitatively assess the collision risk of multiple UAV systems based on the safety loss model and conditional risk value;

[0010] The third step is to design a coordinated warning and obstacle avoidance maneuver trigger mechanism, taking into account the relative speed and distance between the UAV and the dynamic obstacle, and designing a safety threshold;

[0011] The fourth step is to design a risk-triggered distributed safety controller, which consists of basic formation control items, obstacle avoidance control items, and inter-machine collision avoidance control items.

[0012] Furthermore, the first step is specifically as follows:

[0013] Considering the uncertainty of state estimation of multi-UAV system and the uncertainty of UAV’s perception of external static and / or dynamic obstacles, the estimated UAV system state is transformed into and obstacle location information obtained by the sensor Modeled as a general random vector with a joint probability density function.

[0014] In order to quantitatively evaluate the safety risk of multiple drones, a safety loss model based on relative distance information is constructed to quantitatively characterize the insecurity of drones. The drones and the detected obstacles are approximated as circles with radii of and In addition, an additional safety boundary is introduced and , then the safe distance between multiple drones is described as:

[0015] ,

[0016] The safe distance between the drone and external obstacles is described as:

[0017] ,

[0018] Furthermore, considering the distance between the drone and its neighbors and the distance between the drone and external obstacles, the safety loss model between multiple drones is constructed as follows:

[0019] ,

[0020] The safety loss model between the drone and external obstacles is:

[0021] ,

[0022] In the formula , They represent the relative distances between the j-th UAV and the k-th obstacle and the i-th UAV respectively.

[0023] Ideally, by ensuring and This can ensure that multiple drones avoid collisions. However, due to the uncertainty in the drone's state estimation and the position perception of obstacles, the description of the drone's collision risk based on the above deterministic constraints is too conservative. In view of the uncertainty problem, the conditional risk value is used to quantitatively evaluate the collision risk of multiple drones.

[0024] The second step is as follows:

[0025] Based on the safety loss model constructed in the first step, the collision risk of the multi-UAV system is quantitatively evaluated in combination with the conditional risk value. The conditional risk value of the safety loss between multiple UAVs can be calculated by the following formula:

[0026] ,

[0027] Similarly, the conditional risk value between the drone and the obstacle can be calculated as follows:

[0028] ,

[0029] In the formula , z is the risk value, Based on the above formula, the collision risk of multi-UAV system can be quantitatively evaluated.

[0030] The third step is as follows:

[0031] Considering the relative speed, direction and relative distance between the drone and the dynamic obstacle, the inter-drone safety threshold is designed as shown in the following formula:

[0032] ,

[0033] Similarly, the safety threshold between the drone and the obstacle is designed to be:

[0034] ,

[0035] In the formula , is a positive weight coefficient, represents the relative speed between the i-th UAV and the j-th UAV, represents the relative speed between the i-th UAV and the k-th dynamic obstacle, , , They represent the movement speeds of the i-th UAV, the j-th UAV and the k-th dynamic obstacle respectively.

[0036] Based on the collision risk quantitative assessment results in the second step and the above safety thresholds, the idea of ​​event triggering mechanism is introduced to design the following multi-UAV collaborative warning mechanism:

[0037] ,

[0038] in , is the set of all neighboring nodes of the i-th drone. The above formula indicates that once the risk condition of the j-th drone Once triggered, even if the jth UAV itself cannot perceive the obstacle information, it can immediately receive the directional warning from its neighboring nodes, thereby obtaining the relevant status information of the kth obstacle and completing the obstacle avoidance maneuver.

[0039] The fourth step is as follows:

[0040] The safety controller includes basic formation control items, aircraft collision avoidance control items and obstacle avoidance control items. First, the basic formation control items are shown as follows:

[0041] ,

[0042] In the formula and is the gain matrix, and is the relative position parameter of the UAV in the formation, represents the expected trajectory, and is the weight coefficient of the communication topology graph between multiple UAVs.

[0043] Based on the risk quantification assessment results of the second step and the safety threshold of the third step, the risk-triggered multi-UAV inter-UAV collision avoidance control items are designed as follows:

[0044] ,

[0045] In the formula and is the gain matrix.

[0046] The design of multi-UAV obstacle avoidance control items based on risk triggering is as follows:

[0047] ,

[0048] In the formula , is the gain matrix, and is a positive gain coefficient, is a rotation matrix defined as:

[0049] ,

[0050] in is a rotation angle given by:

[0051] ,

[0052] in Indicates the sensing distance of the drone.

[0053] Finally, the basic formation control item, the inter-machine collision avoidance control item and the obstacle avoidance control item are combined as the drone safety controller:

[0054] ,

[0055] In the formula , as well as is the weight coefficient, represents the set of all obstacles faced by the i-th UAV, including the obstacles perceived by itself and the obstacles warned by its neighboring nodes.

[0056] The beneficial effects of the present invention compared with the prior art are:

[0057] The present invention is mainly aimed at the UAV formation system in a dense static and / or dynamic obstacle environment. Compared with the traditional UAV formation obstacle avoidance control method, the present invention can quantitatively evaluate the obstacle risk information in the environment, and then realize the coordinated warning between multiple UAVs through the risk trigger mechanism, thereby breaking through the limitations of the individual perception ability of the UAV, realizing the safe avoidance of dynamic obstacles outside the individual perception range of the UAV, and increasing the safety of the UAV formation in a dense obstacle environment. In addition, the design of the safety controller based on risk triggering can effectively reduce the maneuvering distance of the UAV formation obstacle avoidance, while reducing conservatism. The present invention first considers the uncertainty of system state estimation and obstacle perception, and constructs a safety loss model between the UAV and the external static and / or dynamic obstacles, and between multiple UAVs; then, based on the safety loss model and the conditional risk value, the collision risk of the multi-UAV system is quantitatively evaluated; secondly, a coordinated warning and obstacle avoidance maneuver triggering mechanism is designed, and the relative speed and distance of the UAV and the dynamic obstacle movement are considered, and the safety threshold is designed; finally, a distributed safety controller based on risk triggering is designed, which is composed of a basic formation control item, an obstacle avoidance control item, and an inter-machine collision avoidance control item. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is a schematic diagram of a multi-UAV collaborative early warning and safety control method based on risk triggering of the present invention. DETAILED DESCRIPTION

[0059] Taking a general type of rotor UAV formation as an example, a multi-UAV coordinated early warning and safety control method based on risk triggering of the present invention is described. During the UAV formation execution process, static and / or dynamic obstacles in the environment will pose a threat to the safety of the UAVs. The method proposed by the present invention can ensure the safe flight of the UAV formation system in an environment with dense static and / or dynamic obstacles.

[0060] like Figure 1 As shown, a multi-UAV collaborative early warning and safety control method based on risk triggering of the present invention realizes collaborative early warning and safe obstacle avoidance flight in dense static and / or dynamic obstacle environments through the UAV's own perception ability, risk assessment based on conditional risk value, safety control based on risk triggering, and interaction of inter-machine communication network, including the following steps:

[0061] In the first step, considering the uncertainty of system state estimation and obstacle perception, a safety loss model between UAVs and external static and / or dynamic obstacles, as well as between multiple UAVs, is constructed.

[0062] The second step is to quantitatively assess the collision risk of multiple UAV systems based on the safety loss model and conditional risk value;

[0063] The third step is to design a coordinated warning and obstacle avoidance maneuver trigger mechanism, taking into account the relative speed and distance between the UAV and the dynamic obstacle, and designing a safety threshold;

[0064] The fourth step is to design a risk-triggered distributed safety controller, which consists of basic formation control items, obstacle avoidance control items, and inter-machine collision avoidance control items.

[0065] Furthermore, the first step is specifically as follows:

[0066] Considering the uncertainty of state estimation of multi-UAV system and the uncertainty of UAV’s perception of external static and / or dynamic obstacles, the estimated UAV system state is transformed into and obstacle location information obtained by the sensor Modeled as a general random vector with a joint probability density function.

[0067] In order to quantitatively evaluate the safety risk of multiple drones, a safety loss model based on relative distance information is constructed to quantitatively characterize the insecurity of drones. The drones and the detected obstacles are approximated as circles with radii of and In addition, an additional safety boundary is introduced and , then the safe distance between multiple drones is described as:

[0068] ,

[0069] The safe distance between the drone and external obstacles is described as:

[0070] ,

[0071] Furthermore, considering the distance between the drone and its neighbors and the distance between the drone and external obstacles, the safety loss model between multiple drones is constructed as follows:

[0072] ,

[0073] The safety loss model between the drone and external obstacles is:

[0074] ,

[0075] In the formula , They represent the relative distances between the j-th UAV and the k-th obstacle and the i-th UAV respectively.

[0076] Ideally, by ensuring and This can ensure that multiple drones avoid collisions. However, due to the uncertainty in the drone's state estimation and the position perception of obstacles, the description of the drone's collision risk based on the above deterministic constraints is too conservative. In view of the uncertainty problem, the conditional risk value is used to quantitatively evaluate the collision risk of multiple drones.

[0077] The second step is as follows:

[0078] Based on the safety loss model constructed in the first step, the collision risk of the multi-UAV system is quantitatively evaluated in combination with the conditional risk value. The conditional risk value of the safety loss between multiple UAVs can be calculated by the following formula:

[0079] ,

[0080] Similarly, the conditional risk value between the drone and the obstacle can be calculated as follows:

[0081] ,

[0082] In the formula , z is the risk value, Based on the above formula, the collision risk of multi-UAV system can be quantitatively evaluated.

[0083] The third step is as follows:

[0084] Considering the relative speed, direction and relative distance between the drone and the dynamic obstacle, the inter-drone safety threshold is designed as shown in the following formula:

[0085] ,

[0086] Similarly, the safety threshold between the drone and the obstacle is designed to be:

[0087] ,

[0088] In the formula , is a positive weight coefficient, represents the relative speed between the i-th UAV and the j-th UAV, represents the relative speed between the i-th UAV and the k-th dynamic obstacle, , , They represent the movement speeds of the i-th UAV, the j-th UAV and the k-th dynamic obstacle respectively.

[0089] Based on the collision risk quantitative assessment results in the second step and the above safety thresholds, the idea of ​​event triggering mechanism is introduced to design the following multi-UAV collaborative warning mechanism:

[0090] ,

[0091] in , is the set of all neighboring nodes of the i-th drone. The above formula indicates that once the risk condition of the j-th drone Once triggered, even if the jth UAV itself cannot perceive the obstacle information, it can immediately receive the directional warning from its neighboring nodes, thereby obtaining the relevant status information of the kth obstacle and completing the obstacle avoidance maneuver.

[0092] The fourth step is as follows:

[0093] The safety controller includes basic formation control items, aircraft collision avoidance control items and obstacle avoidance control items. First, the basic formation control items are shown as follows:

[0094] ,

[0095] In the formula and is the gain matrix, and is the relative position parameter of the UAV in the formation, represents the expected trajectory, and is the weight coefficient of the communication topology graph between multiple UAVs.

[0096] Based on the risk quantification assessment results of the second step and the safety threshold of the third step, the risk-triggered multi-UAV inter-UAV collision avoidance control items are designed as follows:

[0097] ,

[0098] In the formula and is the gain matrix.

[0099] The design of multi-UAV obstacle avoidance control items based on risk triggering is as follows:

[0100] ,

[0101] In the formula , is the gain matrix, and is a positive gain coefficient, is a rotation matrix defined as:

[0102] ,

[0103] in is a rotation angle given by:

[0104] ,

[0105] in Indicates the sensing distance of the drone.

[0106] Finally, the basic formation control item, the inter-machine collision avoidance control item and the obstacle avoidance control item are combined as the drone safety controller:

[0107] ,

[0108] In the formula , as well as is the weight coefficient, represents the set of all obstacles faced by the i-th UAV, including the obstacles perceived by itself and the obstacles warned by its neighboring nodes.

[0109] The contents not described in detail in the specification of the present invention belong to the prior art known to the professional and technical personnel in this field.

Claims

1. A multi-UAV collaborative early warning and safety control method based on risk triggering, characterized in that: The following steps are involved: In the first step, considering the uncertainty of system state estimation and obstacle perception, a safety loss model between UAVs and external static and / or dynamic obstacles, as well as between multiple UAVs, is constructed. The second step is to quantitatively assess the collision risk of multiple UAV systems based on the safety loss model and conditional risk value; The third step is to introduce an event trigger mechanism based on the collision risk quantitative assessment results and safety thresholds in the second step; design a coordinated warning and obstacle avoidance maneuver trigger mechanism, consider the relative speed and distance between the UAV and the dynamic obstacle, and design a safety threshold; The fourth step is to design risk-triggered multi-UAV inter-UAV collision avoidance control items based on the risk quantification assessment results of the second step and the safety threshold of the third step; design a risk-triggered distributed safety controller, which consists of basic formation control items, obstacle avoidance control items and inter-UAV collision avoidance control items; The first step is as follows: Considering the uncertainty of state estimation of multi-UAV system and the uncertainty of UAV’s perception of external static and / or dynamic obstacles, the estimated UAV system state is transformed into and obstacle location information obtained by sensors Modeled as a general random vector with a joint probability density function; In order to quantitatively evaluate the safety risk of multiple drones, a safety loss model based on relative distance information is constructed to quantitatively characterize the insecurity of drones. The drones and detected obstacles are approximated as circles with radii of and In addition, an additional safety boundary is introduced and , then the safe distance between multiple drones is described as: ， The safe distance between the drone and external obstacles is described as: ， Considering the distance between a drone and its neighbors and the distance between the drone and external obstacles, the safety loss model between multiple drones is constructed as follows: ， The safety loss model between the drone and external obstacles is: ， In the formula , They represent the relative distances between the j-th UAV and the k-th obstacle and the i-th UAV respectively; Ideally, by ensuring and This can ensure that multiple drones avoid collisions; However, due to the uncertainty of the drone's state estimation and the position perception of obstacles, the description of the drone's collision risk based on the above deterministic constraints is too conservative. To address the uncertainty problem, the conditional risk value is used to quantitatively evaluate the collision risk of multiple drones. The second step is as follows: Based on the safety loss model constructed in the first step, the collision risk of the multi-UAV system is quantitatively evaluated in combination with the conditional risk value; the conditional risk value of the safety loss between multiple UAVs is calculated by the following formula: ， The conditional risk value between the drone and the obstacle is calculated as follows: ， In the formula , z is the risk value, For confidence.

2. The multi-UAV collaborative early warning and safety control method based on risk triggering according to claim 1 is characterized in that: The third step is as follows: Considering the relative speed, direction and relative distance between the drone and the dynamic obstacle, the inter-drone safety threshold is designed as shown in the following formula: ， The safety threshold between the drone and the obstacle is designed to be: ， In the formula , is a positive weight coefficient, represents the relative speed between the i-th UAV and the j-th UAV, represents the relative speed between the i-th UAV and the k-th dynamic obstacle, , , They represent the movement speeds of the i-th UAV, the j-th UAV, and the k-th dynamic obstacle respectively; Based on the collision risk quantitative assessment results in the second step and the above safety thresholds, the idea of ​​event triggering mechanism is introduced to design the following multi-UAV collaborative warning mechanism: ， in , is the set of all neighbor nodes of the i-th drone; The above formula indicates that once the risk condition of the jth drone is Once triggered, even if the jth UAV itself cannot perceive the obstacle information, it can immediately receive the directional warning from its neighboring nodes, thereby obtaining the relevant status information of the kth obstacle and completing the obstacle avoidance maneuver.

3. The multi-UAV collaborative early warning and safety control method based on risk triggering according to claim 1 is characterized in that: The fourth step is as follows: The safety controller includes basic formation control items, inter-machine collision avoidance control items, and obstacle avoidance control items. First, the basic formation control items are shown as follows: ， In the formula and is the gain matrix, and is the relative position parameter of the UAV in the formation, represents the expected trajectory, and is the weight coefficient of the communication topology graph between multiple UAVs; Based on the risk quantification assessment results of the second step and the safety threshold of the third step, the risk-triggered multi-UAV inter-UAV collision avoidance control items are designed as follows: ， In the formula and is the gain matrix; The design of multi-UAV obstacle avoidance control items based on risk triggering is as follows: ， In the formula , is the gain matrix, and is a positive gain coefficient, is a rotation matrix defined as: ， in is a rotation angle given by: ， in Indicates the sensing distance of the drone; Finally, the basic formation control item, the inter-machine collision avoidance control item and the obstacle avoidance control item are combined as the drone safety controller: ， In the formula , as well as is the weight coefficient, represents the set of all obstacles faced by the i-th UAV, including the obstacles perceived by itself and the obstacles warned by its neighboring nodes.

Images