UAV Airborne Collision Avoidance Logic Design Method

Abstract

This invention provides a method for designing airborne collision avoidance logic for unmanned aerial vehicles (UAVs), comprising the following steps: Step 1: Setting initial parameters for the collision avoidance logic; Step 2: Establishing a dynamic model of aircraft encounters based on the initial parameters; Step 3: Calculating the state and corresponding probability of the aircraft at the next moment after performing each action, based on the set turning rate and the dynamic model of aircraft encounters; Step 4: Discretizing the state at the next moment in Step 3 to each discrete point in the state space, calculating the transition probability, and combining the corresponding probability and the transition probability to obtain the final discrete state transition probability; Step 5: Calculating the cost of the aircraft performing each action at the current moment based on the results in Step 4; Step 6: Selecting the optimal action based on the cost of each action in Step 5; Step 7: Repeating Steps 3 to 6 to obtain a complete collision avoidance logic strategy through dynamic programming.

Description

Technical Field

[0001] This invention relates to the field of airborne integrated collision avoidance technology, and more specifically to an airborne collision avoidance logic design method for unmanned aerial vehicles (UAVs). Background Technology

[0002] With the rapid development of drone technology, its applications are becoming increasingly widespread across various fields. However, drones cannot rely on pilots to avoid the risk of collisions with other aircraft in the airspace. To reduce the interference of drone flight activities with other aircraft and ensure flight safety in the airspace, drones need to possess autonomous collision avoidance capabilities. The Traffic Collision Avoidance System (TCAS) currently installed on civil airliners has been proven to significantly reduce the risk of mid-air collisions. However, due to the complex design of its collision avoidance logic, which includes many heuristic rules, it is difficult to modify it to meet the collision avoidance requirements of drones. Summary of the Invention

[0003] In view of this, the embodiments of this specification provide a method for designing airborne collision avoidance logic for unmanned aerial vehicles (UAVs) to improve the reliability and practicality of the collision avoidance system.

[0004] This specification provides the following technical solution in its embodiments: a method for designing airborne collision avoidance logic for unmanned aerial vehicles (UAVs), comprising the following steps: Step 1: setting initial parameters for the collision avoidance logic; Step 2: establishing a dynamic model of aircraft encounters based on the initial parameters for the collision avoidance logic; Step 3: calculating the state of the aircraft at the next moment after each action and the corresponding probability based on the set turning rate and the dynamic model of aircraft encounters; Step 4: discretizing the state at the next moment in Step 3 to each discrete point in the state space and calculating the transition probability, combining the corresponding probability and the transition probability to obtain the final discrete state transition probability; Step 5: calculating the cost of the aircraft performing each action at the current moment based on the results in Step 4; Step 6: selecting the optimal action based on the cost of each action in Step 5; Step 7: repeating Steps 3 to 6 to obtain a complete collision avoidance logic strategy through dynamic programming.

[0005] Furthermore, step two specifically involves the following state during the aircraft encounter process: The discrete dynamic model is: The equation of motion of an aircraft

[0006] for: ;in, It is a random variable representing the horizontal turn rate noise of an aircraft. It refers to the action being performed on this machine at the current moment. The horizontal distance between the two machines. This is the azimuth angle of the local machine relative to the intrusion machine. The relative heading angle of the intruding aircraft. and The speeds of the local machine and the intrusion machine are respectively. and These are the turning speeds of the local machine and the intrusion machine, respectively. This represents the time required for the two aircraft to reach a vertical distance of 0. and These represent the velocities in the x and y directions of the local machine and the intrusion machine, respectively.

[0007] Furthermore, step three specifically involves: using a finite number of sigma sampling points to represent the discrete state of the aircraft. The turning rate after each action in the action set is calculated, and the state at the next moment can be obtained based on these turning rates. ;

[0008] The probability corresponding to this state at that moment is: Where N is the number of sampling points.

[0009] Furthermore, step four specifically involves using multilinear interpolation to calculate the possible states for the next time step. Transition to each discrete state point in the discrete state space Above, and calculate its transition probability as This allows us to determine that the aircraft is in a discrete state. Execute action at time The next time step transitions to a discrete state. The probability is .

[0010] Furthermore, step five specifically involves: in discrete states Execute action The cost is

[0011] ,in, The fixed cost is calculated based on the type of action performed and whether a near-space collision has occurred.

[0012] Furthermore, step six specifically involves: in discrete states The optimal expected cost of performing each action is

[0013] ;

[0014] Then in discrete state The optimal collision avoidance strategy satisfies .

[0015] Furthermore, the UAV airborne collision avoidance logic design method also includes step eight: compressing the logic data from step seven.

[0016] Compared with the prior art, the beneficial effects that can be achieved by the above-mentioned at least one technical solution adopted in the embodiments of this specification include: designing the UAV airborne collision avoidance logic unit by combining decision theory and dynamic programming, and compressing the huge collision avoidance logic table data to improve the reliability and practicality of the collision avoidance system. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating an embodiment of the present invention;

[0019] Figure 2 This is a schematic diagram of an aircraft encounter model in an embodiment of the present invention. Detailed Implementation

[0020] The embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0021] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0022] like Figure 1 and Figure 2 As shown, this embodiment of the invention provides a method for designing airborne collision avoidance logic for unmanned aerial vehicles (UAVs), including the following steps:

[0023] Step 1: Set initial parameters for collision avoidance logic. Based on aircraft performance, airspace requirements, and collision avoidance requirements, set the initial parameters for the collision avoidance logic, including the action set, state dimension, and corresponding discrete state space. The action set includes five actions: Collision-Free Risk (COC), Left Turn (WL), Right Turn (WR), Enhanced Left Turn (SL), and Enhanced Right Turn (SR). The discrete state space has 7 dimensions. ρ, θ, φ, v own 、v int τ and a prev .

[0024] Step 2: Establish a dynamic model of the aircraft encounter. The states during the aircraft encounter process are as follows:

[0025] (1)

[0026] The discrete dynamic model is

[0027] (2)

[0028] in It is a random variable representing the horizontal turn rate noise of an aircraft. This refers to the action being performed by the aircraft at the current moment. The aircraft's equation of motion. It can be represented as

[0029] (3)

[0030] in The horizontal distance between the two machines. This is the azimuth angle of the local machine relative to the intrusion machine. The relative heading angle of the intruding aircraft. and The speeds of the local machine and the intrusion machine are respectively. and These are the turning speeds of the local machine and the intrusion machine, respectively. This is the time required for the two aircraft to reach a vertical distance of 0. , , and The velocities in the x and y directions of the local machine and the intrusion machine, respectively, can be calculated based on the parameters mentioned above.

[0031] Step 3: Calculate the aircraft's state and corresponding probability at the next moment after each action. A finite number of sigma sampling points are used to represent the aircraft's discrete state. The turning rate after each action in the action set is executed, and the state at the next moment can be obtained based on these turning rates.

[0032] (4)

[0033] The probability corresponding to this state is

[0034] (5)

[0035] in N This represents the number of sampling points.

[0036] Sigma sampling of the aircraft's turn rate after each maneuver yields three possible turn rates with probabilities of [0.25, 0.5, 0.5]. The possible turn rates for executing COC are [-1.5, 0, 1.5]° / s, WL is [1.0, 1.5, 2.0]° / s, WR is [-2.0, -1.5, -1.0]° / s, SL is [2.0, 3.0, 4.0]° / s, and SR is [-4.0, -3.0, -2.0]° / s. Combining the possible turn rates of the local and intruder aircraft yields nine turn rate pairs. and the corresponding probability for each turning rate pair Based on these turning rates, we can obtain multiple possible states and corresponding probabilities for the two aircraft to meet in the next moment.

[0037] Step 4: Discretize the possible states of the aircraft at the next time step. Use multilinear interpolation to determine the possible states at the next time step. Transition to each discrete state point in the discrete state space Up. And calculate its transition probability as It can be concluded that the aircraft is in a discrete state. Execute action at time The next time step transitions to a discrete state. The probability is

[0038] (6)

[0039] In one-dimensional linear interpolation, assuming in the interval China x By interpolation, we can obtain the following formula.

[0040]

[0041] in, and The discrete state after interpolation and The probability of the two machines meeting. Each dimension is linearly interpolated into the corresponding interval of the discrete state space, resulting in multiple discrete state points. and their corresponding probabilities are Furthermore, it can be calculated that the aircraft is in a discrete state. Execute action at time The next time step transitions to a discrete state. probability .

[0042] Step 5: Calculate the cost of each action performed by the aircraft at the current moment. (In discrete state) Execute action a The cost is

[0043] (7)

[0044] in The fixed cost is calculated based on the type of action performed and whether a near-space collision has occurred. It is divided into two parts, one part is The fixed cost when =0 depends on the distance between the two machines. Calculate, when When a near-space collision occurs, its value is considered to be 1; otherwise, it decreases with increasing distance. Another part is based on... and The strength was obtained through comparison. The final state satisfies...

[0045]

[0046] Therefore, we can calculate the time steps before the final state. .

[0047] Step 6: Dynamic programming yields the complete collision avoidance logic strategy. In discrete states... The optimal expected cost of performing each action is

[0048] (8)

[0049] Then in discrete state The optimal collision avoidance strategy satisfies

[0050] (9)

[0051] The calculation shows that the machine is in a discrete state. Optimal collision avoidance strategy That is, to execute Minimal action This allows us to obtain the first... t The optimal collision avoidance strategy for each step.

[0052] Step 7: Repeat steps 3-6 until... This allows us to construct the optimal collision avoidance logic strategy for the entire encounter process.

[0053] Step 8: Compress the anti-collision logic data. The original logic data can be divided into 5... 5 = 25 small pieces Each piece has Six dimensions. Subtract each data point from its respective... After finding the minimum value in the direction, it can be observed that there is a certain similarity among many blocks of data, that is...

[0054] (10)

[0055] (11)

[0056] when

[0057] (12)

[0058] Then it is believed They are similar, so they are grouped into an equivalence class. yes standard deviation yes The difference between the maximum and minimum values. After data compression, this can be obtained. m equivalence classes and a minimum value table This allows for storing anti-collision logic strategies in less space, improving the practicality of the anti-collision logic.

[0059] The above description is merely a specific embodiment of the present invention and should not be construed as limiting the scope of the invention. Therefore, any substitution of equivalent components or equivalent changes and modifications made within the scope of protection of this patent should still fall within the scope of this patent. Furthermore, the technical features, technical features and technical solutions, and technical solutions in this invention can be freely combined and used.

Claims

1. A method for designing airborne collision avoidance logic for unmanned aerial vehicles (UAVs), characterized in that, Includes the following steps: Step 1: Set the initial parameters for the anti-collision logic; Step 2: Establish a dynamic model of aircraft encounters based on the initial parameters of the collision avoidance logic; Step 3: Based on the set turn rate and the aircraft encounter dynamic model, calculate the aircraft's state and corresponding probability at the next moment after each action; Step 4: Discretize the state at the next time step in Step 3 to each discrete point in the state space, calculate the transition probability, and combine the corresponding probability and the transition probability to obtain the final discrete state transition probability. Step 5: Based on the results of Step 4, calculate the cost of each action performed by the aircraft at the current moment; Step Six: Select the optimal action based on the cost value of each action in Step Five; Step 7: Repeat steps 3 to 6 to obtain a complete collision avoidance logic strategy through dynamic programming; Step two specifically involves: The state of the aircraft during the encounter is as follows: ; The discrete dynamic model is: ; Aircraft equations of motion for: ； in, It is a random variable representing the horizontal turn rate noise of an aircraft. It refers to the action being performed on this machine at the current moment. The horizontal distance between the two machines. This is the azimuth angle of the local machine relative to the intrusion machine. The relative heading angle of the intruding aircraft. and The speeds of the local machine and the intrusion machine are respectively. and These are the turning speeds of the local machine and the intrusion machine, respectively. This represents the time required for the two aircraft to reach a vertical distance of 0. and These represent the velocities in the x and y directions of the local machine and the intrusion machine, respectively.

2. The UAV airborne collision avoidance logic design method according to claim 1, characterized in that, Step three specifically involves: using a finite number of sigma sampling points to represent the discrete state of the aircraft. The turning rate after each action in the action set is used to obtain the state at the next moment, based on these turning rates: ; The probability corresponding to this state at that moment is: Where N is the number of sampling points.

3. The UAV airborne collision avoidance logic design method according to claim 2, characterized in that, Step four specifically involves: using multilinear interpolation to determine the possible states for the next time step. Transition to each discrete state point in the discrete state space Above, and calculate its transition probability as The aircraft is in a discrete state. Execute action at time The next time step transitions to a discrete state. The probability is .

4. The UAV airborne collision avoidance logic design method according to claim 3, characterized in that, Step five specifically involves: in the discrete state Execute action The cost is ,in, The fixed cost is calculated based on the type of action performed and whether a near-space collision has occurred.

5. The UAV airborne collision avoidance logic design method according to claim 4, characterized in that, Step six specifically involves: in the discrete state The optimal expected cost of performing each action is ； Then in discrete state The optimal collision avoidance strategy satisfies .

6. The UAV airborne collision avoidance logic design method according to any one of claims 1 to 5, characterized in that, The UAV airborne collision avoidance logic design method further includes step eight: compressing the logic data in step seven.

Images