A legged robot cluster hunting control method based on interaction matrix

Abstract

A kind of multi-legged robot cluster hunting control method based on interaction matrix, main steps include: constructing multi-legged robot cluster system dynamics model;Set the cluster hunting control task and target of different task stages;Establish the interaction matrix model of different task stages;Design the cluster hunting control strategy based on interaction matrix;In the process of executing task, cluster robot carries out dynamic role assignment;Based on the above strategy, the multi-legged robot gradually completes the formation of the expected hunting formation from the initial position, and then completes the hunting tracking and avoids collision with obstacles. This method can realize high-robustness adaptive cooperative control by dynamically modeling the interaction force between robots and the terrain interaction force, and combining with the dynamic role assignment mechanism, effectively improves the terrain adaptability of the cluster in unstructured environment, and provides a key technical path for large-scale application in complex scenarios.

Description

Technical Field

[0001] This invention relates to the field of legged robot swarm control technology, and in particular to a swarm encirclement control method based on an interaction force matrix. Background Technology

[0002] Legged robots, with their excellent adaptability to complex terrain and high mobility, have shown broad application prospects in disaster relief, field exploration, and military reconnaissance. However, single legged robots suffer from low efficiency and limited coverage when performing large-scale tasks. Swarm control technology, which enables efficient task allocation and collaborative execution through multi-robot cooperation, has become a current research hotspot.

[0003] Complex terrain requires the encirclement formation to have high real-time performance and adaptability. However, traditional methods have low collaborative efficiency and insufficient obstacle avoidance ability because they do not fully model the dynamic mechanical coupling relationship between robots. At the same time, fixed role allocation strategies (such as leader-follower models) are prone to uneven load (such as the leader consuming more than 40% more energy than the follower), rigid function switching and single point of failure risk, making it difficult to cope with dynamic task requirements. Summary of the Invention

[0004] This disclosure provides a method for controlling the encirclement of legged robot swarms based on an interaction force matrix. By dynamically modeling the interaction forces between robots and the terrain forces, and combining it with a dynamic role allocation mechanism, it achieves highly robust adaptive cooperative control, significantly improving the swarm's terrain adaptability in unstructured environments, and providing a key technical path for large-scale applications in complex scenarios.

[0005] The method for controlling the swarming of legged robots based on the interaction force matrix disclosed herein mainly includes the following steps:

[0006] S1, Construct a dynamic model of a multi-legged robot cluster system;

[0007] S2, set the cluster capture and control tasks and objectives for different task stages;

[0008] S3, Establish interaction force matrix models for different task stages;

[0009] S4, Design a cluster capture control strategy based on the interaction force matrix;

[0010] S5: During the task execution process, the cluster robots dynamically assign roles.

[0011] Based on the above method, the legged robot gradually completes the formation of the desired encirclement formation from the initial position, thereby completing the encirclement and tracking while avoiding collisions with obstacles.

[0012] Furthermore, the specific method of step S1 includes:

[0013] Suppose that the cluster system contains n legged robots, denoted by N = {1, 2, 3, ..., n}, where N i Let be the set of neighbors of the i-th legged robot;

[0014] Using the leader-follower method, n legged robots are divided into 1 leader and n-1 followers, denoted by L and F respectively;

[0015] Let the safety range of the legged robot be s. i Referring to Newton's second law, the kinematic model of the i-th legged robot is as follows:

[0016]

[0017] Where p i (t)=(x i (t),y i (t) represents the position of the i-th legged robot, x i y i Let v represent the horizontal and vertical coordinates of the i-th legged robot, respectively. i θ represents the velocity of the legged robot. i (t), φ i (t) represent the heading angle and turning rate of the i-th legged robot, respectively;

[0018] Based on the dynamic characteristics of legged robots, the kinematic constraints are as follows:

[0019]

[0020] Where v m φ m a m These represent the magnitudes of maximum speed, maximum turning rate, and maximum acceleration, respectively.

[0021] Furthermore, the specific method of step S2 includes:

[0022] The encirclement and control task is divided into two phases: encirclement formation and encirclement tracking. The tasks and objectives of each phase include:

[0023] S21, Formation stage of the encirclement and capture:

[0024] The tasks in this phase include: adjusting multiple legged robots from an initial dispersed or randomly distributed state to a formation with a desired geometry and spatial structure by forming a control strategy of encirclement and formation;

[0025] The objective of this phase is to determine the relative positions and distances between the legged robots based on the requirements of the encirclement mission, forming a formation that can effectively surround the target. Specific mission objectives include:

[0026] (1) All legged robots maintain the desired encirclement formation, i.e., the relative distance between the legged robots is a given c = {c ij}, where c ij This represents the expected distance between the i-th legged robot and the j-th legged robot;

[0027] (2) Each legged robot has a safety range s during the formation process. i To prevent collisions between legged robots and with other obstacles;

[0028] S22, Encirclement and Tracking Phase:

[0029] The tasks in this phase include: after the encirclement formation is formed, the legged robot adjusts its position in real time according to the target's state through an encirclement and tracking control strategy to maintain the encirclement of the target and ultimately achieve the capture of the target.

[0030] The objective of this phase is to dynamically adapt the encirclement formation to the target's movement and changes, surrounding the target within the formation. Specific objectives include:

[0031] (1) After the encirclement formation is established, the legged robot system tracks the specific target p as a whole. r If the legged robot i is the leader, then ||p i -p r ||Converges to 0;

[0032] (2) For rigid formations, legged robots move at the same speed, that is, the heading angle and speed of adjacent robots are consistent;

[0033] (3) During the tracking process, the target of the encirclement formation stage should be maintained at the same time.

[0034] Furthermore, the specific method of step S3 includes:

[0035] Let r be the sensing range of the i-th legged robot. i Then the set of neighbors of the i-th legged robot that can be perceived is represented as N. i ={j|d ij ≤r i};

[0036] Let the actual position of the legged robot i be p. i (t), the actual position of the legged robot j is p. j (t), then:

[0037] d ij (t)=||p i (t)-p j (t)||

[0038] Where ||·|| is the vector 2 norm, d ij (t) represents the distance between the i-th legged robot and the j-th legged robot at time t;

[0039] The specific methods for constructing the interaction force matrix model for different task stages include:

[0040] S31, Formation stage of the encirclement and capture formation

[0041] During the formation phase of the encirclement, the force analysis of the i-th legged robot is as follows:

[0042]

[0043] Among them, F i,formation (t) represents the interaction force F experienced by the legged robot i at time t during the formation phase of the encirclement formation. i,j (t) represents the attractive or repulsive force of legged robot j on legged robot i, M i,m (t) represents the repulsive force F exerted by obstacle m on legged robot i. f (t) represents the ground friction force. A collection of obstacles;

[0044] Specifically:

[0045] (1)F i,j (t) is constructed as follows:

[0046] F i,j (t)=k n (c ij -d ij (t))u ij (t)

[0047] Among them, 0 <k n <1 is an adjustment parameter, u ij (t)=(p i (t)-p j (t)) / d ij (t);

[0048] To avoid singular problems, let d ij When (t) < δ, F i,j (t)=k n u0; where δ is a pre-given sufficiently small number, and u0 is a random unit vector;

[0049] When dij (t) <c ij When the actual distance is less than the expected distance, legged robot j exerts a repulsive force on legged robot i.

[0050] When d ij (t)>c ij When the actual distance is greater than the expected distance, legged robot j attracts legged robot i.

[0051] d ij (t)=c ij At that time, legged robot j does not exert force on legged robot i;

[0052] (2)M i,m (t) is constructed as follows:

[0053]

[0054] Among them, 0 <k m <1 is an adjustment parameter, u im (t)=(p i (t)-p m ) / d im (t), p m Let m be the position of the m-th obstacle;

[0055] To avoid singular problems, let d im When (t) < δ, M i,m (t)=k m u0, where δ is a pre-given sufficiently small number, and u0 is a random unit vector;

[0056] When d im (t) i At that time, the actual distance between the legged robot i and the obstacle m is less than the safe distance of the legged robot, and the obstacle m exerts a repulsive force on the legged robot i;

[0057] (3)F f (t) is constructed as follows:

[0058] F f (t)=μ(v i (t))·F n ·sign(v i (t))

[0059] in, ∈=0.001m / s;

[0060] The μ(v) function is as follows:

[0061] ​

[0062] Where, μ s μ is the static friction coefficient. k v is the coefficient of kinetic friction. s Stribeck characteristic velocity;

[0063] F n The normal force is as follows:

[0064]

[0065] Where, m i Let N be the mass of the i-th legged robot, g be the acceleration due to gravity, and N be the mass of the i-th legged robot. contact This represents the number of feet currently touching the ground.

[0066] The interaction force matrix for the formation stage of the encirclement formation is:

[0067] F formation (t)=[F i,formation [t)|i∈N];

[0068] S32, Encirclement and Tracking Phase

[0069] During the encirclement and tracking phase, force analysis is performed on the i-th legged robot:

[0070]

[0071] Among them, F i,tracking (t) represents the interactive forces experienced by the legged robot i during the capture and tracking phase at time t, where L and F are the leader set and follower set, respectively. i,p (t) represents the attraction of the target point to the legged robot, and is constructed as follows:

[0072]

[0073] Among them, 0 <k p <1 is an adjustment parameter, p r (t) represents the position of the target point at time t; to avoid singular problems, assume that when ||p i (t)-p r When (t)||<δ, F i,p (t)=-k p u0, where δ is a pre-given sufficiently small number, and u0 is a random unit vector;

[0074] The interaction force matrix model for the capture and tracking phase is then:

[0075] F tracking (t)=[F i,tracking (t)|i∈N].

[0076] Furthermore, the specific method of step S4 includes:

[0077] Based on the potential field method framework, the legged robot system dynamically constructs an interaction force matrix and adjusts parameters at different task stages, enabling the robot to autonomously adjust its speed and heading angle under the action of interaction forces, ultimately achieving efficient swarm capture of targets; specific strategies include:

[0078] S41, Formation stage of the encirclement and capture formation

[0079] In the interaction force F i,formation Under the influence of (t), the speed and direction of the i-th legged robot change. Simultaneously, limited by its mobility, the speed at time t+1 changes by Δv compared to time t. i And the change in heading angle Δφ i for:

[0080] Δv i =min(a m ,|F i,formation (t)|-v i (t))

[0081]

[0082] Among them, v i (t) represents the current velocity of the i-th legged robot at time t, a m For its maximum acceleration, w m Its maximum angular velocity, φ i (t) is its heading angle at time t;

[0083] S42, Encirclement and Tracking Phase

[0084] In the interaction force F i,tracking Under the influence of (t), the speed and direction of the i-th legged robot change. Simultaneously, limited by its mobility, the speed at time t+1 changes by Δv compared to time t. i And the change in heading angle Δφ i for:

[0085] Δv i =min(a m ,|F i,tracking (t)|-v i (t))

[0086]

[0087] Among them, v i (t) represents the current velocity of the i-th legged robot at time t, a m For its maximum acceleration, w m Its maximum angular velocity, φi (t) is its heading angle at time t.

[0088] Furthermore, the specific method of step S5 includes:

[0089] The leader periodically broadcasts a heartbeat signal σ(t), as follows:

[0090]

[0091] Leader switching includes two modes: passive switching and active switching, wherein:

[0092] The failure of a legged robot results in a passive switchover. If a follower does not receive a heartbeat within a time window T, the leader is deemed to have failed, where T = 3 × T. heart ,T heart The heartbeat cycle;

[0093] When a follower becomes more suitable to lead the current leader during formation tracking, a proactive switch is initiated, involving the following steps:

[0094] The leader calculates the Euclidean distance d between himself and the dynamic target in real time. ir (t)=p i (t)-p r (t), the follower calculates its own Euclidean distance d to the dynamic target in real time. jr (t)=p j (t)-p r (t);

[0095] If d jr (t) <d ir (t)-Δd th This triggers a handover request, where Δd th The switching threshold;

[0096] Each legged robot calculates its own leadership ability function W. i The one with the highest ability score becomes the new leader:

[0097]

[0098] Where α, β, and γ are weighting coefficients, and d i Let E be the distance between the i-th legged robot and the target. i Let C be the normalized power of the i-th legged robot. i To determine the communication quality of the i-th legged robot after normalization, C i Specifically as follows:

[0099]

[0100] Among them, SNR max SNR min Let SNR be the maximum and minimum values ​​of SNR in legged robots. i Let be the signal-to-noise ratio of the i-th legged robot, specifically as follows:

[0101]

[0102] For the i-th legged robot, P i,signal To receive signal power, P i,noise This represents the power of environmental noise.

[0103] This disclosure constructs an interactive force model that integrates foot contact force, robot-to-robot interaction force, and terrain action force. By combining real-time environmental perception data, it dynamically coordinates the mechanical coupling relationship between individuals within the cluster, thereby significantly improving motion robustness and collaborative task execution efficiency in complex terrain.

[0104] Compared with the prior art, the beneficial effects of this disclosure are:

[0105] ① This method achieves multi-level coupling of mechanical model calculation, behavioral decision-making and control execution, and can dynamically adjust the role of the legged robot while avoiding collisions and maintaining formation stability;

[0106] ② Strong environmental adaptability and flexible and controllable formation: Through real-time interactive force matrix modeling and dynamic role allocation mechanism, it can adaptively adjust the formation shape and control strategy according to terrain features and task requirements, and meet the requirements for flexible obstacle avoidance;

[0107] ③ Outstanding stability and significant anti-interference capability: By constructing an interactive force model that integrates foot contact force, robot interaction force and terrain action force, the individual movement can be coordinated in real time in a dynamic environment. Even when there is a sudden external force interference (such as a single robot stalling), the cluster can still quickly reconstruct the formation through the interactive force matrix to complete the encirclement task. Attached Figure Description

[0108] The above and other objects, features and advantages of this disclosure will become more apparent from the more detailed description of exemplary embodiments of this disclosure taken in conjunction with the accompanying drawings, in which the same reference numerals generally represent the same components.

[0109] Figure 1 This is a general block diagram of a swarm control system for encircling and trapping a legged robot according to the present disclosure;

[0110] Figure 2 A schematic diagram of the dynamics modeling of a legged robot system;

[0111] Figure 3 Force analysis for legged robots (encirclement formation stage);

[0112] Figure 4 Force analysis for legged robots (encirclement and tracking phase);

[0113] Figure 5 This is a schematic diagram of the hexagonal formation tracking process. Detailed Implementation

[0114] Preferred embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While preferred embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the present disclosure will be thorough and complete, and will fully convey the scope of the present disclosure to those skilled in the art.

[0115] This disclosure proposes a swarm control method for legged robots based on an interaction force matrix. By establishing a multi-dimensional interaction force matrix, the dynamic coupling relationship of forces between swarm members and between members and the environment is described. Based on the interaction force matrix, a distributed cooperative control framework is constructed to drive the swarm to achieve adaptive cooperative capture and robust motion control in complex terrain.

[0116] As attached Figure 1 As shown, in one exemplary implementation, the main steps of the method include: constructing a dynamic model of a multi-legged robot system, setting a swarm capture control task objective, establishing an interaction force matrix model, designing a swarm capture control strategy, and dynamically assigning roles.

[0117] 1. Dynamics Modeling of Multi-legged Robot Systems

[0118] Suppose that the cluster system contains n legged robots, denoted by N = {1, 2, 3, ..., n}. i Let L be the set of neighbors of the i-th legged robot. Using the leader-follower method, the n legged robots are divided into one leader and n-1 followers, denoted by L and F respectively. Figure 2 As shown, the safe range of the legged robot is s. i Referring to Newton's second law, the kinematic model of the i-th legged robot is shown below:

[0119]

[0120] Where p i (t)=(x i (t),y i (t) represents the position of the i-th legged robot, x i y i Let v represent the horizontal and vertical coordinates of the i-th legged robot, respectively. iθ represents the velocity of the legged robot. i (t), φ i (t) represent the heading angle and turning rate of the i-th legged robot, respectively.

[0121] Based on the dynamic characteristics of legged robots, the kinematic constraints are as follows:

[0122]

[0123] Where v m φ m a m These represent the magnitudes of maximum speed, maximum turning rate, and maximum acceleration, respectively.

[0124] 2. Set up cluster capture and control task objectives

[0125] The encirclement and control task can be divided into two phases: encirclement formation and encirclement tracking. Considering the position, velocity, and turning rate of the legged robot, the objectives for each phase can be obtained:

[0126] Encirclement formation formation stage:

[0127] Encirclement formation refers to the process by which multiple legged robots, initially dispersed or randomly distributed, adjust themselves into a formation with a desired geometry and spatial structure through specific control strategies and algorithms. This process involves position adjustment, obstacle avoidance, and cooperative path planning to ensure the formation of an encirclement area for the legged robots. The main objective of this stage is to determine the relative positions and distances between the legged robots according to the requirements of the encirclement task, forming a formation that effectively surrounds the target. Specific task objectives are as follows:

[0128] (1) All legged robots should maintain the desired encirclement formation, i.e., the relative distance between legged robots should be a given c = {c ij}, where c ij This represents the expected distance between the i-th legged robot and the j-th legged robot;

[0129] (2) Each legged robot needs to consider a safety range s during the formation process. i This prevents collisions between legged robots and with other obstacles.

[0130] Encirclement and tracking phase:

[0131] Encirclement and tracking refers to the process where, after forming an encirclement formation, legged robots adjust their positions in real time according to the target's state to maintain an encirclement and ultimately capture the target. The main objective of this stage is to enable the encirclement formation to dynamically adapt to changes in the target's movement, keeping the target within the formation. This process requires dynamically adjusting the positions of each legged robot to cope with external disturbances, environmental changes, or path adjustments, while maintaining the stability of the encirclement formation.

[0132] (1) After the encirclement formation is established, the legged robot system needs to track the specific target p as a whole. r If the legged robot i is the leader, then ||p i -p r || It should converge to 0;

[0133] (2) For rigid formations, legged robots should move at the same speed, that is, the heading angle and speed of adjacent robots should be consistent;

[0134] (3) The tracking process still requires maintaining the formation, so the objectives of the encirclement formation stage also need to be considered in this stage.

[0135] 3. Establish an interaction force matrix model

[0136] The sensing range of the i-th legged robot is r i Then the set of neighbors of the i-th legged robot that can be perceived is represented as N. i ={j|d ij ≤r i The actual position of the legged robot i is p. i (t), the actual position of the legged robot j is p. j (t), then:

[0137] d ij (t)=||p i (t)-p j (t)||

[0138] Where ||·|| is the vector 2 norm, d ij (t) represents the distance between the i-th legged robot and the j-th legged robot at time t.

[0139] (1) Formation of the encirclement formation stage

[0140] During the formation phase of the encirclement, a force analysis is performed on the i-th legged robot:

[0141]

[0142] Among them, F i,formation(t) represents the interaction force F experienced by the legged robot i at time t during the formation phase of the encirclement formation. i,j (t) represents the attractive or repulsive force of legged robot j on legged robot i, M i,m (t) represents the repulsive force F exerted by obstacle m on legged robot i. f (t) represents the ground friction force. A collection of obstacles.

[0143] Specifically, F i,j (t) is constructed as follows:

[0144] F i,j (t)=k n (c ij -d ij (t))u ij (t)

[0145] Among them, 0 <k n <1 is an adjustment parameter, u ij (t)=(p i (t)-p k (t)) / d ij (t). To avoid singular problems, when d ik When (t) < δ, F i,k (t)=k n u0. Where δ is a pre-given sufficiently small number, and u0 is a random unit vector. When d ij (t) <c ij When the actual distance is less than the expected distance, legged robot j exerts a repulsive force on legged robot i; when d ik (t)>c ij When the actual distance is greater than the expected distance, legged robot j exerts an attractive force on legged robot i; d ij (t)=c ij At that time, legged robot j does not exert force on legged robot i.

[0146] M i,m (t) is constructed as follows:

[0147]

[0148] Among them, 0 <k m <1 is an adjustment parameter, u im (t)=(p i (t)-p m ) / d im (t), p m Let be the position of the m-th obstacle. To avoid singularities, when d im When (t) < δ, M i,m(t)=k m u0. Where δ is a pre-given sufficiently small number, and u0 is a random unit vector. When d im (t) i At that time, the actual distance between the legged robot i and the obstacle m is less than the safe distance of the legged robot, and the obstacle m exerts a repulsive force on the legged robot i.

[0149] F f (t) is constructed as follows:

[0150] F f (t)=μ(v i (t))·F n ·sign(v i (t))

[0151] in, ∈=0.001m / s.

[0152] The μ(v) function is as follows:

[0153]

[0154] Where, μ s μ is the static friction coefficient. k v is the coefficient of kinetic friction. s This is the characteristic velocity of Stribeck.

[0155] F n The normal force is as follows:

[0156]

[0157] Where, m i Let N be the mass of the i-th legged robot, g be the acceleration due to gravity, and N be the mass of the i-th legged robot. contact This represents the number of feet currently touching the ground.

[0158] Force analysis during the formation of the encirclement formation, as follows: Figure 3 As shown, its interaction force matrix is ​​modeled as follows:

[0159] F formation (t)=[F i,formation (t)|i∈N]

[0160] (2) Encirclement and tracking phase

[0161] During the encirclement and tracking phase, force analysis is performed on the i-th legged robot:

[0162]

[0163] Among them, F i,tracking ​(t) represents the interactive forces experienced by the legged robot i during the capture and tracking phase at time t, where L and F are the leader set and follower set, respectively. i,p (t) represents the attraction of the target point to the legged robot, and is constructed as follows:

[0164]

[0165] Among them, 0 <k p <1 is an adjustment parameter, p r (t) represents the position of the target point at time t. To avoid singularity issues, when ||p i (t)-p r When (t)||<δ, F i,p (t)=-k p u0. Where δ is a pre-given sufficiently small number, and u0 is a random unit vector.

[0166] Force analysis during the encirclement and tracking phase, as follows Figure 4 As shown, its interaction force matrix is ​​modeled as follows:

[0167] F tracking (t)=[F i,tracking (t)|i∈N]

[0168] 4. Design a cluster capture and control strategy

[0169] Based on the potential field method framework, the motion of legged robots in the environment is designed as motion within an abstract gravitational field. Target points exert an attractive force on the legged robots, while obstacles exert a repulsive force. The interaction forces between robots and the terrain forces are also considered. Finally, the motion of the legged robots is controlled by calculating the resultant force. Therefore, the legged robot system dynamically constructs an interaction force matrix and adjusts parameters at different task stages, enabling the robots to autonomously adjust their speed and heading angle under the influence of interaction forces, ultimately achieving efficient swarm capture of targets.

[0170] (1) Formation of the encirclement formation stage

[0171] In the interaction force F i,formation Under the influence of (t), the speed and direction of the i-th legged robot change. Simultaneously, limited by its mobility, the speed at time t+1 changes by Δν compared to time t. i And the change in heading angle Δφ i for:

[0172] Δν i =min(a m ,|F i,formation (t)|-ν i (t))

[0173]

[0174] Where, ν i (t) represents the current velocity of the i-th legged robot at time t, a m For its maximum acceleration, w m For its maximum angular velocity, δ i (t) is its heading angle at time t;

[0175] (2) Encirclement and tracking phase

[0176] In the interaction force F i,tracking Under the influence of (t), the speed and direction of the i-th legged robot change. Simultaneously, limited by its mobility, the speed at time t+1 changes by Δν compared to time t. i And the change in heading angle Δφ i for:

[0177] Δν i =min(a m ,|F i,tracking (t)|-ν i (t))

[0178]

[0179] Where, ν i (t) represents the current velocity of the i-th legged robot at time t, a m For its maximum acceleration, w m Its maximum angular velocity, φ i (t) is its heading angle at time t;

[0180] 5. Dynamic role allocation mechanism

[0181] In leader-follower formation control, dynamic role allocation mechanisms are a key technology for improving system robustness, adaptability, and efficiency. Its core objective is to dynamically adjust the leader's role based on environmental changes, mission requirements, or member status (such as insufficient energy or malfunctions) without interrupting the encirclement mission, thereby ensuring formation stability and mission continuity.

[0182] The leader periodically broadcasts a heartbeat signal σ(t), as follows:

[0183]

[0184] Leader switching occurs in two modes: passive switching and active switching. Passive switching occurs when the legged robot fails; if a follower does not receive a heartbeat within a time window T, the leader is considered to have failed. Where T = 3 × T heart T heart This refers to the heartbeat cycle.

[0185] When a follower becomes more suitable to lead the current leader during formation tracking, a proactive switch occurs. The leader calculates the Euclidean distance d between itself and the dynamic target in real time. ir (t)=p i (t)-p r (t), the follower calculates its own Euclidean distance d to the dynamic target in real time. jr (t)=p j (t)-p r (t), if d jr (t) <d ir (t)-Δd th This triggers a switchover request. Where Δd th =0.5m, to avoid frequent changes in leadership.

[0186] Each legged robot calculates its own leadership ability function W. i Those with higher ability scores will become the new leaders.

[0187]

[0188] Where α, β, and γ are weighting coefficients, and d i Let E be the distance between the i-th legged robot and the target. i Let C be the normalized power of the i-th legged robot. i Let C be the communication quality of the i-th legged robot after normalization. i Specifically as follows:

[0189]

[0190] Among them, SNR max SNR min Let SNR be the maximum and minimum values ​​of SNR in legged robots. i Let be the signal-to-noise ratio of the i-th legged robot, specifically as follows:

[0191]

[0192] For the i-th legged robot, P i,signal To receive signal power, P i,noise For ambient noise power, SNR i The higher the value, the stronger the communication reliability.

[0193] 6. Simulation Experiment

[0194] like Figure 5As shown, the swarm system comprises six legged robots, consisting of one leader and five followers. The safe range for each legged robot is 0.3m, and the desired encirclement formation is a regular hexagon. This formation ensures that the target is evenly surrounded in all directions, preventing escape from any direction. The leader and follower legged robots are represented by blue triangles and red dots, respectively. Green pentagons represent target points that the formation needs to track, and blue circles represent obstacles.

[0195] Based on this cluster formation control method, the legged robot initially has an irregular shape. It gradually completes the first stage, forming the desired encirclement formation shape, and then completes the second stage, performing encirclement and tracking while avoiding collisions with obstacles.

[0196] The legged robot swarm control method based on the interaction force matrix disclosed herein has initially solved the problems of formation instability and rigid role allocation in complex terrain, and provides a reliable and adaptive swarm control solution for high dynamic scenarios such as military collaborative reconnaissance and disaster relief.

[0197] The above technical solutions are merely exemplary embodiments of the present invention. For those skilled in the art, based on the application methods and principles disclosed in the present invention, it is easy to make various types of improvements or modifications, and not limited to the methods described in the specific embodiments of the present invention. Therefore, the methods described above are merely preferred and not restrictive.

Claims

1. A method for controlling the swarming of legged robots based on an interaction force matrix, characterized in that, Includes the following steps: S1, Construct a dynamic model of a multi-legged robot cluster system; S2, set the cluster capture and control tasks and objectives for different task stages; S3, Establish the interaction force matrix model for different task stages; S4, Design a cluster capture control strategy based on the interaction force matrix; S5: During the execution of tasks, the cluster robots dynamically assign roles. Based on the above method, the legged robot gradually completes the formation of the desired encirclement formation from the initial position, thereby completing the encirclement and tracking while avoiding collisions with obstacles; The specific methods for constructing the interaction force matrix model for different task stages include: S31, Formation stage of the encirclement and capture formation During the formation of the encirclement formation, the first i The force analysis of the legged robot is as follows: in, express Legged robots i The interactive forces experienced during the formation phase of the encirclement formation. representing legged robots j Legged robots i attraction or repulsion Indicates obstacles m Legged robots i The repulsive force, This represents the frictional force on the ground. A collection of obstacles; The interaction force matrix for the formation phase of the encirclement formation is: ； S32, Encirclement and Tracking Phase During the encirclement and tracking phase, for the first i Force analysis of a legged robot: in, express Legged robots i The interactive forces experienced during the encirclement and tracking phase. and They are the leader set and the follower set, The attraction of the target point to the legged robot is represented by the following structure: in, To adjust the parameters, For legged robots i Actual location for t The position of the target point at any given time; to avoid singularities, assume that when hour, ,in, Given a sufficiently small number, It is a random unit vector; The interaction force matrix model for the capture and tracking phase is then: 。 2. The method according to claim 1, characterized in that, The specific method of step S1 includes: Suppose the cluster system contains n A legged robot, using It means that, among them, For the first i A collection of legged robot neighbors; Using the leader-follower approach, n The legged robot is divided into one leader and one... Each follower, using and express; Let the safety range of the legged robot be Referring to Newton's second law, the... i The kinematic model of the legged robot is shown below: in Indicates the first i The position of the legged robot , They represent the first i The horizontal and vertical coordinates of a legged robot This indicates the speed of the legged robot. , They represent the first i The heading angle and turning rate of a legged robot; Based on the dynamic characteristics of legged robots, the kinematic constraints are as follows: in , These represent the magnitudes of maximum speed, maximum turning rate, and maximum acceleration, respectively.

3. The method according to claim 1, characterized in that, The specific method of step S2 includes: The encirclement and control task is divided into two phases: encirclement formation and encirclement tracking. The tasks and objectives of each phase include: S21, Formation stage of the encirclement and capture: The tasks in this phase include: adjusting multiple legged robots from an initial dispersed or randomly distributed state to a formation with a desired geometry and spatial structure by forming a control strategy of encirclement and formation; The objective of this phase is to determine the relative positions and distances between the legged robots based on the requirements of the encirclement mission, forming a formation that can effectively surround the target. Specific mission objectives include: (1) All legged robots maintain the desired encirclement formation, i.e., the relative distance between the legged robots is given. ,in This represents the expected distance between the i-th legged robot and the j-th legged robot; (2) Each legged robot has a safety range during the formation process. To prevent collisions between legged robots and with other obstacles; S22, Encirclement and Tracking Phase: The tasks in this phase include: after the encirclement formation is formed, the legged robot adjusts its position in real time according to the target's state through an encirclement and tracking control strategy to maintain the encirclement of the target and ultimately achieve the capture of the target. The objective of this phase is to dynamically adapt the encirclement formation to the target's movement and changes, surrounding the target within the formation. Specific objectives include: (1) After the encirclement formation is established, the legged robot system tracks the specific target as a whole. If legged robots i As a leader, Converging to 0; (2) For rigid formations, legged robots move at the same speed, that is, the heading angle and speed of adjacent robots are consistent; (3) Maintain the target during the encirclement formation phase while tracking.

4. The method according to claim 3, characterized in that, The specific method of step S3 includes: Let the first i The sensing range of a single-legged robot is Then the first i The set of neighbors of a legged robot that can perceive other legged robots is represented as follows: ; Legged robot i The actual location is Legged robots j The actual location is ,but: in, For vector 2 norm, for t Time of the first i A legged robot and the first j The distance between individual legged robots; Specifically: (1) The structure is as follows: in, To adjust the parameters, ; To avoid strange problems, assume that when hour, ;in, Given a sufficiently small number, It is a random unit vector; when When the actual distance is less than the expected distance, the legged robot... j Legged robots i Generates repulsive force; when When the actual distance is greater than the expected distance, the legged robot... j Legged robots i Generate attraction; At that time, legged robots j non-legged robots i Generate force; (2) The structure is as follows: in, To adjust the parameters, = , For the first m The location of the obstacle; To avoid strange problems, assume that when hour, ,in, Given a sufficiently small number, It is a random unit vector; when At that time, legged robots i and obstacles m The actual distance between them is less than the safe distance for legged robots, obstacles m Legged robots i Generates repulsive force; (3) The structure is as follows: in, ; The function is as follows: in, The static friction coefficient is The coefficient of kinetic friction is . Stribeck characteristic velocity; The normal force is as follows: in, For the first i The quality of a legged robot It is the acceleration due to gravity. This represents the number of feet currently touching the ground.

5. The method according to claim 4, characterized in that, The specific method of step S4 includes: Based on the potential field method framework, the legged robot system dynamically constructs an interaction force matrix and adjusts parameters at different task stages, enabling the robot to autonomously adjust its speed and heading angle under the action of interaction forces, ultimately achieving efficient swarm capture of targets; specific strategies include: S41, Formation stage of the encirclement and capture formation In interaction Under the influence of the first i The speed and direction of movement of a legged robot change, but it is limited by its movement capabilities. Compared to time Change in speed at any moment for: in, for Time of the first i The current speed of the legged robot Its maximum acceleration, Its maximum angular velocity, For Heading angle at any moment; S42, Encirclement and Tracking Phase In interaction Under the influence of the first i The speed and direction of movement of a legged robot change, but it is limited by its movement capabilities. Compared to time Change in speed at any moment for: in, for Time of the first i The current speed of the legged robot Its maximum acceleration, Its maximum angular velocity, For Heading angle at any time.

6. The method according to any one of claims 1-5, characterized in that, The specific method of step S5 includes: Leader periodically broadcasts heartbeat signals (t), as follows: Leader switching includes two modes: passive switching and active switching, wherein: The failure of a legged robot results in a passive switchover. If a follower does not receive a heartbeat within a time window T, the leader is deemed to have failed. , The heartbeat cycle; When a follower becomes more suitable to lead the current leader during formation tracking, a proactive switch is initiated, involving the following steps: Leaders calculate the Euclidean distance between themselves and dynamic targets in real time. The follower calculates its Euclidean distance to the dynamic target in real time. ; like This triggers a switch request, where... The switching threshold; Each legged robot calculates its own leadership ability function. The one with the highest ability score becomes the new leader: in, These are the weighting coefficients. For the first i The distance between the legged robot and the target. For the remainder after normalization, the th i The battery capacity of a legged robot After normalization, the first i Communication quality of legged robots Specifically as follows: in, , These represent the maximum and minimum SNR values ​​for legged robots. Let be the signal-to-noise ratio of the i-th legged robot, specifically as follows: Among them, for the first i Legged robot To receive signal power, This represents the power of environmental noise.

Images