Method and device for path planning of swarm intelligent robot based on virtual potential field

Abstract

The application discloses a kind of group intelligence robot path planning method and device based on virtual potential field, and wherein, method includes: the multiple virtual potential field forces of hunting robot and hunting object when executing hunting task are calculated;After the multiple virtual potential field forces of hunting robot are superimposed, the movement speed of hunting robot is calculated, the multiple virtual potential field forces of hunting object are superimposed to calculate the movement speed of hunting object, and the movement of hunting robot and hunting object is controlled in real time according to the movement speed of hunting robot and the movement speed of hunting object.By doing so, on the basis of the advantages of traditional virtual force field, the movement accuracy, sensitivity, reaction speed of robot can be further improved, and has good expandability, and efficiently completes more complex antagonistic game task.

Description

Technical Field

[0001] This application relates to the field of path planning technology, and in particular to a path planning method and apparatus for swarm intelligent robots based on a virtual potential field. Background Technology

[0002] Existing path planning algorithms in the field of swarm intelligence still have many shortcomings. For example, ant colony optimization and artificial bee colony optimization are highly dependent on parameters, easily get trapped in local optima, and require a long time to develop positive feedback, resulting in slow convergence. Particle swarm optimization algorithms suffer from low accuracy and convergence difficulties when particle velocities are high. In summary, current path planning algorithms for swarm intelligent robots struggle to simultaneously meet the requirements of high robustness, decentralization, easy scalability, and strong adaptability, and are also difficult to implement in adversarial game tasks that demand high robot efficiency. Summary of the Invention

[0003] This application provides a path planning method and device for swarm intelligent robots based on virtual potential fields. On the basis of the advantages of traditional virtual force fields, it can further improve the robot's motion accuracy, sensitivity and reaction speed, and has good scalability, enabling it to efficiently complete more complex adversarial game tasks.

[0004] The first aspect of this application provides a path planning method for swarm intelligent robots based on a virtual potential field, comprising the following steps: calculating a first attractive force between the swarming robot and the swarming object based on the predicted position and predicted velocity of the swarming object at the next moment; calculating a first repulsive force between the swarming robots based on distance information between the swarming robots; calculating a second repulsive force between the swarming robots based on distance information between the swarming robots and obstacles; calculating a loosening force between the swarming robots based on distance information and position information between the swarming robots; and calculating a third repulsive force of the swarming robot on the swarming object based on the positions of the swarming object, the swarming robot, and the obstacles. The system calculates the encircling force of the robot when the object is stationary or when the distance between the robot and the object is less than a preset value. It then calculates the first attractive force, the first repulsive force, the second repulsive force, the loosening force, and the encircling force to calculate the robot's speed. Finally, it calculates the object's speed by superimposing the third, fourth, and fifth repulsive forces. The system then controls the movement of both the robot and the object in real time based on their respective speeds.

[0005] Optionally, in one embodiment of this application, before calculating the first attraction between the trapping robot and the trapping object, the method further includes: predicting the predicted position of the trapping object at the next moment using a Kalman filter; comparing the predicted position at the next moment with the actual position of the trapping object at the next moment; updating the state matrix and covariance matrix of the Kalman filter based on the comparison result; and using the updated Kalman filter to predict the predicted position and predicted velocity of the trapping object at the next moment.

[0006] Optionally, in one embodiment of this application, the first attraction F between the trapping robot and the trapping object... g for:

[0007]

[0008] Where k1 and k2 are both positive proportionality coefficients, R g r is the radius of the enclosing circle. g Let r be the vector pointing from the coordinates of the robot to the coordinates of the target being captured. g For vector r g modulus, v g The absolute speed of the target is calculated based on the predicted speed of the target and the speed of the target robot itself.

[0009] Optionally, in one embodiment of this application, the first repulsive force F between the encircling robots c for:

[0010]

[0011] Where, r g Let r be the vector pointing from the coordinates of the robot to the coordinates of the target being captured. g For vector r g The modulus, R s To determine the safe radius between the robots, r h Let r be the vector pointing from the coordinates of the capturing robot to the coordinates of the evading capturing robot. h For vector r h The modulus, k5 is the proportionality coefficient.

[0012] Optionally, in one embodiment of this application, the surrounding force F of the trapping robot r for:

[0013]

[0014] Where k4 is a direct proportionality coefficient.

[0015] Optionally, in one embodiment of this application, the loosening force F of the containment robotl for:

[0016]

[0017] τ=(-sinθ g cosθ g )

[0018] Where, θ g F represents the azimuth of the target relative to itself. l Let τ be the magnitude of the loosening force, τ be the tangential unit vector of the target relative to itself, and smooth be the smoothing function.

[0019] Optionally, in one embodiment of this application, when the number of the trapping robots within the field of vision of the target is less than or equal to 2, the third repulsive force F of the trapping robots on the target is... h for:

[0020]

[0021] Where k6 is a direct proportionality coefficient, r h r is a vector pointing from the coordinates of the target to the coordinates of the capturing robot. h For vector r h The modulus;

[0022] When the number of the trapping robots within the field of vision of the target is greater than 2, the third repulsive force F exerted by the trapping robots on the target is... h for:

[0023]

[0024] Among them, R e The warning distance between the target and the capturing robot is denoted by h, which is a unit vector pointing from the capturing robot to the incenter of the triangle formed by all the capturing robots.

[0025] Optionally, in one embodiment of this application, the fifth repulsive force F of the target itself... e for:

[0026]

[0027] Where k7 is a proportionality coefficient.

[0028] Optionally, in one embodiment of this application, after calculating the movement speed of the trapping robot and the movement speed of the target, the method further includes: correcting the movement speed of the trapping robot and the movement speed of the target using an incremental PD algorithm.

[0029] A second aspect of this application provides a path planning device for swarm intelligent robots based on a virtual potential field, comprising: a first calculation module for calculating a first attractive force between the swarming robot and the swarming object based on the predicted position and predicted velocity of the swarming object at the next moment; a second calculation module for calculating a first repulsive force between the swarming robots based on distance information between the swarming robots, and calculating a second repulsive force between the swarming robots based on distance information between the swarming robots and an obstacle; a third calculation module for calculating a loosening force of the swarming robots based on distance information and position information between the swarming robots; and a fourth calculation module for calculating the swarming robot's position based on the position of the swarming object, the swarming robot, and the obstacle. The system calculates the encircling force of the robot when the target is stationary or when the distance between the robot and the target is less than a preset value. The planning module is used to superimpose the first attractive force, the first repulsive force, the second repulsive force, the loosening force, and the encircling force to calculate the speed of the robot; it also superimposes the third, fourth, and fifth repulsive forces to calculate the speed of the target; and controls the movement of the robot and the target in real time based on their respective speeds.

[0030] The virtual potential field-based path planning method and apparatus for swarm intelligent robots in this application enable swarm intelligent robots to move quickly, accurately, and robustly without a leader. While achieving intelligent obstacle avoidance, they also function as an organic whole to efficiently and stably complete adversarial game tasks (such as surrounding or attacking a target robot). It is highly scalable and has low hardware requirements; only the same independent program needs to be written into the robots, and each robot only needs to collect local information to join the team and complete the task together.

[0031] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0032] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0033] Figure 1 This is a flowchart of a swarm intelligent robot path planning method based on a virtual potential field, according to an embodiment of this application.

[0034] Figure 2This is a schematic diagram comparing the hunting efficiency of a hunting robot under different conditions according to an embodiment of this application;

[0035] Figure 3 This is a schematic diagram comparing the impact of adding a speed correction term on the hunting efficiency of a robot according to the embodiments of this application;

[0036] Figure 4 This is a schematic diagram of the smooth function provided according to an embodiment of this application;

[0037] Figure 5 This is a schematic diagram of the loosening force action provided according to an embodiment of this application;

[0038] Figure 6 This is a schematic diagram illustrating the escape of a target when applying the improved force field model according to an embodiment of this application.

[0039] Figure 7 This is a schematic diagram of the newly introduced additional repulsive field provided according to embodiments of this application;

[0040] Figure 8 This is an example diagram of a swarm intelligent robot path planning device based on a virtual potential field according to an embodiment of this application. Detailed Implementation

[0041] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0042] Figure 1 This is a flowchart of a swarm intelligent robot path planning method based on a virtual potential field, according to an embodiment of this application.

[0043] like Figure 1 As shown, the path planning method for swarm intelligent robots based on virtual potential fields includes the following steps:

[0044] In step S101, the first attraction between the trapping robot and the trapping object is calculated based on the predicted position and predicted velocity of the trapping object at the next moment.

[0045] Optionally, in one embodiment of this application, before calculating the first attraction between the trapping robot and the trapping object, the method further includes: using a Kalman filter to predict the predicted position of the trapping object at the next moment; comparing the predicted position at the next moment with the actual position of the trapping object at the next moment; updating the state matrix and covariance matrix of the Kalman filter based on the comparison result; and using the updated Kalman filter to predict the predicted position and predicted velocity of the trapping object at the next moment.

[0046] Unlike conventional autonomous driving missions, the encirclement robot does not reach the target point when carrying out an encirclement mission. Instead, it forms a circle with the target (prey) as the center and a radius of approximately R. g The encirclement of the target is crucial. In adversarial game tasks, because the target has a certain ability to evade capture, the encircling robot will experience additional deviations due to program computation delays during each location of the target and motion planning. Therefore, during path planning, the encircling robot needs to make corresponding corrections and fine adjustments based on the target's movement speed. However, in actual adversarial situations, the target may have high maneuverability, and its speed information cannot be simply obtained based on its displacement. To obtain a more accurate target's movement speed, the encircling robot can use the Kalman filter method to predict the target's next node position, thereby estimating the target's current movement strategy.

[0047] The Kalman filter consists of two stages: prediction and update. In the prediction stage, the predicted position of the target in the next state can be obtained from the current position information of the target. Based on the requirements, let the state matrix at time k be denoted by x, and the target position vector be denoted by x. k The covariance matrix is ​​P k Kalman filtering assumes that the variables in the state matrix are random and follow a Gaussian distribution. Since noise exists in the system, it needs to be considered during prediction. Therefore, the final prediction equation is as follows:

[0048]

[0049]

[0050] Where F k Q is the state transition matrix, which is the identity matrix according to the kinematic relationship; k The covariance is represented by Q. k The noise.

[0051] After the location of the target changes, the Kalman gain K is introduced to update the state matrix and covariance matrix by comparing the observed actual location with the predicted location for the next prediction. The specific formula is as follows:

[0052]

[0053] P k '=P k -K'H k P k

[0054]

[0055] This model can be used to obtain the predicted position of the next node of the target being surrounded, thereby estimating the target's movement strategy at this time.

[0056] In three scenarios—(1) the speed of the target is unknown, (2) the speed of the target is predicted using Kalman filtering, and (3) the speed of the target is completely known—100 simulation tests were conducted, and the average time was taken. To ensure that the hunting robot can catch up with the target, the hunting robot is set to always move at its maximum speed. The average hunting time in the three scenarios is as follows: Figure 2 As shown, this method for predicting the speed of the hunted object can significantly improve hunting efficiency, and the Kalman filter method's estimation of the hunted object's strategy matches the actual situation well.

[0057] In a hunting mission, the hunting robot does not directly reach the target point; its ultimate task is to maintain a distance around the target and form a stable radius of R. g Therefore, when designing the forces exerted by the target on the trapping robot, a simple attraction model cannot be used. Instead, a virtual force field is designed by simulating intermolecular forces, using r g The vector r represents the coordinates of the robot that is being captured, pointing to the coordinates of the target. g Let k1 be the modulus of the vector, then the design attraction is: (where k1 is a proportionality coefficient).

[0058]

[0059] In adversarial game tasks, the center of the encirclement constantly shifts due to the movement of the target, making the encirclement robot's movement more complex and difficult to form a stable encirclement. Therefore, the traditional potential function based solely on position is not directly applicable; the potential field needs to be modified to account for the velocities of both the encircling robot and the target. In this case, the encircling robot will use the Kalman filter method described above to obtain the predicted relative velocity v of the target. gr And obtain the absolute velocity v of the target based on the speed of the capture robot itself. g By modifying the pseudo-intermolecular force model proposed in the static hunting project, the first attraction F between the hunting robot and the hunting target is obtained. g for:

[0060]

[0061] Where k1 and k2 are positive proportionality coefficients.

[0062] When |r g -R g When the time is sufficiently short, the hunting robot essentially enters the final encirclement phase, needing to closely follow the prey. Since the prey's movement becomes more predictable at this stage, the robot can directly utilize the prey's previous velocity, v. g_back And increase its coefficient term, modifying the attractiveness to (where k1 and k3 are proportionality coefficients):

[0063]

[0064] After conducting 200 simulation tests under the same conditions, adding the above speed correction term during the encirclement process can significantly improve the encirclement efficiency, such as... Figure 3 As shown, it can be expected that the speed correction term will play a more significant role when the prey's speed becomes more complex.

[0065] In step S102, the first repulsive force between the capture robots is calculated based on the distance information between them, and the second repulsive force between the capture robots and the obstacle is calculated based on the distance information between them.

[0066] Unlike the ideal model of the simulation environment, the actual trapping robots have a non-negligible spatial volume. Therefore, in order to avoid collisions between the trapping robots, a safety radius R needs to be set between them. s And it is designed with a certain repulsive force to avoid collisions. Using r h The vector r represents the coordinates of the target being pursued pointing to the coordinates of the target being evaded. h The repulsive force model F is designed for the modulus of this vector. c As shown below: (where k5 is the proportionality coefficient)

[0067]

[0068] When the robot is far away from other objects, the virtual force will not consider the repulsive force, so as to avoid affecting the normal movement of other objects due to the large influence area of ​​its own repulsive force field. However, when the distance enters a certain safe radius, the repulsive force will gradually increase as the distance gradually approaches.

[0069] Since the trapping robot can visually identify obstacles at long distances, the obstacles are incorporated into this virtual force field, and a repulsive force model of the same form is adopted, with the scaling factor appropriately adjusted to r. o r represents the vector pointing from the coordinates of the capture robot to the coordinates of the obstacle. o Let r be the magnitude of this vector, replacing the r mentioned above.h With r h That's all.

[0070] Experiments have shown that the algorithm can effectively avoid collisions between the capture robot and its surrounding environment, and achieves better path planning while taking obstacles into account.

[0071] A significant problem with the virtual force field algorithm is the potential for local optima, where the net potential force on the hunting robot is zero at a single point, rendering it unable to move. In adversarial game tasks, the movement of the hunted object causes more frequent changes in the potential field, making the location of local optima more prone to shift. This uncertainty increases the likelihood that the hunting robot will temporarily become trapped in newly emerging local optima, leading to a decrease in hunting efficiency. To avoid situations where excessive repulsive force prevents the hunting robot from approaching the hunted object during the pursuit, the repulsive force model was modified so that the repulsive force increases with the distance r between the robot and the hunted object. g The repulsive force decreases as the repulsive force decreases. The newly modified repulsive force model is as follows:

[0072]

[0073] This model adds to the previous version. This design reduces the repulsive force between the hunting robots when they are far from the target, preventing the repulsive force from affecting their efficiency in chasing the target. Conversely, when the hunting robots are close to the target, the repulsive force increases, causing the robots to disperse and thus better form an encirclement formation.

[0074] In step S103, the loosening force of the trapping robots is calculated based on the distance and position information between the trapping robots.

[0075] Since hunting missions require the robots to surround the target from all directions, they should move more dispersedly during the encirclement process to avoid premature clustering. To address this, a looseness preference rule is introduced for the hunting robots. This means that the robots tend to move towards looser spaces, which manifests as a loose force. Under this force, the robots are more likely to move into spaces with fewer robots, thus enabling advanced strategies such as flanking maneuvers during the hunt.

[0076] Because the target object is moving, the reference frame established based on the target object is actually moving. Since the model in a static project design inherently lacks continuity, the impact of this problem is amplified when the reference frame begins to move, making the output loose force extremely unstable. To solve this problem, a smoothing function is introduced to correct the output force. This function uses trigonometric functions as transitions to smoothly connect discontinuities, ensuring the stable formation of the encirclement under dynamic hunting. Let `begin` represent the lower limit of the discontinuity and `end` represent the upper limit of the discontinuity. For any input quantity `x`, such as... Figure 4 As shown, the specific mapping relationship of this function is as follows:

[0077]

[0078] While a rigorous mathematical definition of loose space is provided in related technologies, practical testing revealed that achieving the highest efficiency in trapping under its loose preference rule was ultimately difficult. Therefore, certain modifications were made to the loose preference rule. For example... Figure 5 As shown, the capture robot will obtain the positions r1 and r2 of the two nearest friendly robots within its field of vision and the position r of the target object. g A Cartesian coordinate system is established with the robot itself as the origin and the line connecting the robot and the target as the y-axis. The modified model's operating logic is as follows:

[0079] If there are no friendly robots in the field of vision, the looseness force is zero, and the looseness preference rule does not take effect.

[0080] If there is one friendly robot in the field of view, then:

[0081]

[0082] θ e Given the expected angle between adjacent robots when the final encirclement formation is formed, and assuming the number of friendly robots n is known, we have: The effect of this force is to cause the robot to move towards the other side of the target being captured, and to adjust the included angle as much as possible towards θ. e near.

[0083] If there are two or more robots in the field of view, the positions of the two closest robots are obtained. If both friendly robots are on the same side of the target, the loosening force points to the other side, causing the robot to tend to move towards a more open space. If the two friendly robots are on opposite sides of the target, then:

[0084]

[0085] Based on the distances between the two nearest friendly robots and the robot, the force selects the side farther away from the robot as the loose space and tends to move towards this side.

[0086] Loosening forces primarily disperse robots, enabling advanced strategies such as flanking and encirclement. Therefore, they should mainly function when the group is relatively close to the target. Thus, the following modifications are made to minimize their impact on robot movement at greater distances from the target, while maximizing their effectiveness at closer proximity:

[0087] τ=(-sinθ g cosθ g )

[0088]

[0089] Where, θ g F represents the azimuth of the target relative to itself. i Let τ be the magnitude of the loosening force, τ be the tangential unit vector of the target relative to itself, and smooth be the smoothing function.

[0090] After simulation and actual testing, the frequency of motion discontinuity of the robot during the target encirclement process was significantly reduced after the adjustment, which enabled it to capture prey more efficiently and form a regular and stable encirclement more easily.

[0091] In step S104, the third repulsive force of the hunting robot on the hunting object, the fourth repulsive force of the obstacle on the hunting object, and the fifth repulsive force of the hunting object itself are calculated based on the positions of the hunting object, the hunting robot, and the obstacle. When the hunting object is stationary or the distance between the hunting robot and the hunting object is less than a preset value, the surrounding force of the hunting robot is calculated.

[0092] To prevent prey from escaping the surrounding robots and prolong the capture time, a new escape algorithm needs to be developed. Corresponding to the "search-chase-encircle" process of the surrounding robots, the prey itself also undergoes a corresponding "avoidance-escape-breakout" phase, which is closely related to the number and location of the surrounding robots. Therefore, considering the algorithm's practicality and simplicity at each stage, the virtual potential field method is still used in the design of the surrounding robots.

[0093] When the number of hunters in the target's field of vision is less than or equal to 2, it is impossible for the target to completely surround it. However, since new hunting robots may appear at any time, allowing an encirclement to form, the target should still try to stay away from the hunting robots in its field of vision. Because the hunters' movements at this stage only require simple repulsive force guidance, the hunting robot uses a relatively scientific and simple Coulomb force model to calculate the virtual force exerted by the hunting robot on the target. Let r... hThe vector r represents the coordinates of the target being captured pointing to the coordinates of the capturing robot. h Let k be the modulus of this vector, and establish the following mapping relationship between k and k: (where k6 is a proportionality coefficient)

[0094]

[0095] Furthermore, the prey also needs to avoid obstacles, so the same strategy can be used to obtain the repulsive force F exerted by the obstacle on the prey. b .

[0096] When the number of encircling robots exceeds two, a more sophisticated design of the potential field is needed to enhance the adversarial nature of the target and prolong the encirclement period. When the target is outside the polygon with the encircling robots as vertices, the strategy of moving away from the robots remains sufficient. However, once the target enters the polygon, a simple repulsive force model will create a local optimum within the polygon, trapping the target inside. Under this model, the encircling robots only need to perform an encirclement maneuver to prevent the target from escaping, indicating significant room for improvement. Therefore, for two closely spaced encircling robots, the potential field model is modified to an electric dipole model, expressed as follows:

[0097]

[0098] The effect of this potential field model is to make prey try to move away from the hunters furthest away from it, while simultaneously attempting to fight against the two nearest hunters and break out from there, such as... Figure 6 As shown.

[0099] To prevent the target from making incorrect decisions due to the continued attraction between the two nearest capturing robots after the target breaks through, a repulsive field is superimposed on the original force field:

[0100]

[0101] This power field and r h The positive correlation means that when the target is inside the polygon, the repulsive force provided by the field is relatively small, and the electric dipole force dominates, ensuring the successful escape. When the target is outside the polygon, the repulsive field becomes dominant, allowing the target to move significantly away from the robot. Figure 7 As shown.

[0102] After all the encirclement robots have approached sufficiently, in order to achieve complete encirclement, the encirclement robots need to confirm that the target has stopped before they begin to surround it, that is, at v r =0 and |r g -R g When the time is short enough, the hunter itself generates a driving force F that surrounds it.r Its specific expression is: (where k4 is a direct proportionality coefficient)

[0103]

[0104] in In other cases F r =0.

[0105] In step S105, the first attractive force, the first repulsive force, the second repulsive force, the loosening force and the surrounding force are superimposed to calculate the movement speed of the capture robot. The third repulsive force, the fourth repulsive force and the fifth repulsive force are superimposed to calculate the movement speed of the target. The movement of the capture robot and the target is controlled in real time according to the movement speed of the capture robot and the movement speed of the target.

[0106] Optionally, in one embodiment of this application, after calculating the movement speed of the capture robot and the movement speed of the target, the method further includes: correcting the movement speed of the capture robot and the movement speed of the target using an incremental PD algorithm.

[0107] The Raspblock vision omnidirectional vehicle uses a Raspberry Pi 4B as its motherboard, which can connect to cameras and LiDAR for target recognition and positioning, and can also control the motor output of the Mecanum wheels to achieve real-time motion control.

[0108] During a single movement, the capture robot acquires external information through cameras and obstacle avoidance modules, and calculates the pseudo-intermolecular force F. g Hunter repulsion F c Repulsive force F from obstacles o Loosening force F l and surrounding force F r The resultant force F is thus calculated. all for:

[0109]

[0110] By F all The expected position vector v of the capture robot's movement speed was obtained. e =m c ×F all , where m c As a fixed parameter, the speed is further adjusted based on the specific stage the robot is in. If the calculated expected speed is greater than the robot's maximum speed, it needs to be adjusted to the maximum speed. Furthermore, when the target is far away, i.e., the robot is still in the pursuit phase, the robot needs to maintain full speed, denoted as v. max To achieve the maximum output rate of the capture robot, the adjusted speed is: No adjustments are needed when the robot is close to the target.

[0111] Finally, to improve the robustness of the capture robot's motion, an incremental PD algorithm is used to correct the final output velocity v of the capture robot, denoted as v_i of the capture robot's velocity at the previous moment. back Let m and n be the proportional coefficient and differential coefficient, respectively. Then the final output control speed is:

[0112] v=m×v′+n×(v back -v′)

[0113] This speed will control the movement of the capture robot.

[0114] Regarding the movement of the target being pursued, once the encirclement by the robots has been largely formed, the target ceases its escape attempt, and the output velocity v... p =0; For the target being pursued, due to its relatively low average speed, it will most likely need to escape the pursuing robot at full speed. Therefore, the resultant force F of all virtual forces is 0. m Its main function is to provide the predicted direction of motion. The specific calculation method is as follows:

[0115]

[0116] And the expected speed v p The size of ′ is determined by the distance L from the nearest capturing robot to the target. c The decision is made when the nearest capture robot is outside the warning range R. f When the target is stopped, the target will remain stationary; when the nearest target robot is less than the threshold R... n At that time, the target will move at full speed v pmax The transition zones during motion will be connected by a smooth function, the specific expression of which is:

[0117] v p = (1-smooth(L) c ,R n ,R f ))×v pmax

[0118] Combining the two, the prey's expected speed v p 'for:

[0119]

[0120] Similarly, the final output speed of the prey is also corrected using an incremental PD algorithm.

[0121] After multiple experiments, it was verified that the robot can achieve rapid, uniform and stable encirclement and capture under this algorithm, while the prey can also escape the encirclement and capture as much as possible and prolong its survival time.

[0122] The path planning method for swarm intelligent robots based on a virtual potential field proposed in this application has better environmental adaptability. Even with obstacles, it can incorporate them into the virtual force field, achieving intelligent and efficient path planning. The virtual force field is optimized by adding a smoothing function, making the robot's movement smoother. It can handle more complex adversarial game tasks, better fitting real-world application scenarios. When faced with unexpected situations such as members leaving or joining the group, the group can spontaneously adjust its path and complete the task. It can be practically applied to drones, unmanned vehicles, and other carriers to achieve decentralized swarm intelligence, and the robot scale can be expanded through simple replication.

[0123] Next, referring to the accompanying drawings, a swarm intelligent robot path planning device based on a virtual potential field, according to an embodiment of this application, is described.

[0124] Figure 8 This is an example diagram of a swarm intelligent robot path planning device based on a virtual potential field according to an embodiment of this application.

[0125] like Figure 8 As shown, the swarm intelligent robot path planning device 10 based on virtual potential field includes: a first calculation module 100, a second calculation module 200, a third calculation module 300, a fourth calculation module 400, and a planning module 500.

[0126] The system comprises the following modules: a first calculation module 100, used to calculate a first attractive force between the capture robot and the target based on the predicted position and velocity of the target at the next moment; a second calculation module 200, used to calculate a first repulsive force between the capture robots based on the distance information between them, and a second repulsive force between the capture robots and obstacles based on the distance information between them; a third calculation module 300, used to calculate a loosening force between the capture robots based on the distance and position information between them; and a fourth calculation module 400, used to calculate a third repulsive force between the capture robot and the target, a fourth repulsive force between the obstacle and the target, and a fifth repulsive force of the target itself based on the positions of the target, the capture robot, and obstacles, and to calculate the surrounding force of the capture robot when the target is stationary or the distance between the capture robot and the target is less than a preset value. The planning module 500 is used to calculate the movement speed of the capture robot by superimposing the first attraction force, the first repulsion force, the second repulsion force, the loosening force and the surrounding force, and to calculate the movement speed of the target object by superimposing the third repulsion force, the fourth repulsion force and the fifth repulsion force, and to control the movement of the capture robot and the target object in real time according to the movement speed of the capture robot and the movement speed of the target object.

[0127] It should be noted that the foregoing explanation of the embodiment of the path planning method for swarm intelligent robots based on virtual potential field also applies to the path planning device for swarm intelligent robots based on virtual potential field in this embodiment, and will not be repeated here.

[0128] The virtual potential field-based swarm intelligent robot path planning device proposed in this application has better environmental adaptability. In the presence of obstacles, it can incorporate obstacles into the virtual force field, achieving intelligent and efficient path planning. The virtual force field is optimized by adding a smoothing function, making the robot's movement smoother. It can perform more complex adversarial game tasks, better fitting real-world application scenarios. When faced with unexpected situations such as members leaving or joining the group, the group can spontaneously adjust its path and complete the movement task. It can be practically applied to drones, unmanned vehicles, and other carriers to achieve decentralized swarm intelligence, and the robot scale can be expanded through simple replication.

[0129] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0130] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0131] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0132] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0133] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

Claims

1. A path planning method for swarm intelligent robots based on a virtual potential field, characterized in that, Includes the following steps: The first attraction between the trapping robot and the trapping object is calculated based on the predicted position and speed of the target at the next moment. The first repulsive force between the capture robots is calculated based on the distance information between the capture robots, and the second repulsive force between the capture robots and the obstacles is calculated based on the distance information between the capture robots and the obstacles. The loosening force of the encirclement robots is calculated based on the distance and position information between them; The third repulsive force of the hunting robot on the hunting object, the fourth repulsive force of the obstacle on the hunting object, and the fifth repulsive force of the hunting object itself are calculated based on the positions of the hunting object, the hunting robot, and the obstacle. When the hunting object is stationary or the distance between the hunting robot and the hunting object is less than a preset value, the surrounding force of the hunting robot is calculated. Where the number of the trapping robots within the field of vision of the target is less than or equal to 2, the third repulsive force of the trapping robots on the target. for: in, It is a direct proportionality coefficient. The vector pointing from the coordinates of the target to the coordinates of the capturing robot. For vectors The modulus; When the number of the trapping robots within the field of vision of the target is greater than 2, the third repulsive force exerted by the trapping robots on the target. for: in, The warning distance between the target and the robot being apprehended. Let be the unit vector pointing from the trapping robot to the incenter of the triangle formed by all the trapping robots; The movement speed of the trapping robot is calculated by superimposing the first attractive force, the first repulsive force, the second repulsive force, the loosening force, and the surrounding force. The movement speed of the target is calculated by superimposing the third repulsive force, the fourth repulsive force, and the fifth repulsive force. The movement of the trapping robot and the target is controlled in real time based on the movement speed of the trapping robot and the movement speed of the target.

2. The method according to claim 1, characterized in that, Before calculating the first attraction between the trapping robot and the trapping object, the following is also included: The next moment's predicted position of the target is predicted using a Kalman filter. The predicted position at the next moment is compared with the actual position of the target at the next moment. Based on the comparison result, the state matrix and covariance matrix of the Kalman filter are updated, and the updated Kalman filter is used to predict the predicted position and velocity of the target at the next moment.

3. The method according to claim 1, characterized in that, The first attraction between the trapping robot and the target being trapped for: in, and All are positive proportionality coefficients. The radius of the enclosing circle is 1. Let be the vector pointing from the coordinates of the robot to the coordinates of the target being captured. For vectors modulus, The absolute speed of the target is calculated based on the predicted speed of the target and the speed of the target robot itself.

4. The method according to claim 1, characterized in that, The first repulsive force between the encircling robots for: in, Let be the vector pointing from the coordinates of the robot to the coordinates of the target being captured. For vectors modulus, To ensure a safe radius between the robots, Let be the vector pointing from the coordinates of the capturing robot to the coordinates of the evading capturing robot. For vectors modulus, This is the proportionality coefficient.

5. The method according to claim 2, characterized in that, The surrounding force of the encirclement robot for: in, It is a direct proportionality coefficient. .

6. The method according to claim 3, characterized in that, The loosening force of the encirclement robot for: in, The azimuth angle of the target relative to itself. For the magnitude of the loosening force, Let be the tangential unit vector of the target relative to itself. It is a smooth function.

7. The method according to claim 1, characterized in that, The fifth repulsive force of the target itself. for: in, It is a direct proportionality coefficient.

8. The method according to any one of claims 1-7, characterized in that, After calculating the movement speed of the capture robot and the movement speed of the target, the method further includes: The incremental PD algorithm is used to correct the movement speed of the capture robot and the movement speed of the target being captured.

9. A swarm intelligent robot path planning device based on a virtual potential field, characterized in that, include: The first calculation module is used to calculate the first attraction between the trapping robot and the trapping object based on the predicted position and speed of the trapping object at the next moment. The second calculation module is used to calculate the first repulsive force between the capture robots based on the distance information between the capture robots, and to calculate the second repulsive force between the capture robots based on the distance information between the capture robots and the obstacles. The third calculation module is used to calculate the loosening force of the trapping robots based on the distance and position information between them. The fourth calculation module is used to calculate the third repulsive force of the hunting robot on the hunting object, the fourth repulsive force of the obstacle on the hunting object, and the fifth repulsive force of the hunting object itself based on the positions of the hunting object, the hunting robot, and the obstacle. When the hunting object is stationary or the distance between the hunting robot and the hunting object is less than a preset value, the module calculates the surrounding force of the hunting robot. The planning module is used to calculate the movement speed of the trapping robot by superimposing the first attractive force, the first repulsive force, the second repulsive force, the loosening force and the surrounding force, and to calculate the movement speed of the trapping object by superimposing the third repulsive force, the fourth repulsive force and the fifth repulsive force, and to control the movement of the trapping robot and the trapping object in real time according to the movement speed of the trapping robot and the movement speed of the trapping object. Where the number of the trapping robots within the field of vision of the target is less than or equal to 2, the third repulsive force of the trapping robots on the target. for: in, It is a direct proportionality coefficient. The vector pointing from the coordinates of the target to the coordinates of the capturing robot. For vectors The modulus; When the number of the trapping robots within the field of vision of the target is greater than 2, the third repulsive force exerted by the trapping robots on the target. for: in, The warning distance between the target and the robot being apprehended. Let be the unit vector pointing from the trapping robot to the incenter of the triangle formed by all the trapping robots.

Images