Ant colony optimization algorithm-based logistics shuttle vehicle path optimization method and system

Abstract

The application relates to the technical field of logistics automation, and particularly discloses a logistics shuttle path optimization method and system based on an ant colony optimization algorithm, which collects real-time positions of shuttles, track topologies, task start and end points and time stamps, and obtains initial space-time data; an expected driving path is expressed as a continuous time curve, and historical path occupation is mapped as continuous space-time pheromone concentration values; the continuous time curve is taken as an artificial ant colony individual, repulsion potential based on predicted trajectories of other shuttles is introduced, a joint cost is constructed, curve control points are iteratively adjusted, the curve with the minimum joint cost is selected as an optimal continuous path according to conflict integral sorting, pheromone concentration on the path is reversely enhanced, and the optimal continuous path is converted into a speed-time instruction sequence; the application performs path searching in a continuous space-time domain, eliminates discretization deviation, reduces the probability of multi-shuttle conflicts, and realizes adaptive optimization of the path.

Description

Technical Field

[0001] This invention relates to the field of logistics automation technology, specifically to a method and system for optimizing logistics shuttle routes based on ant colony optimization algorithm. Background Technology

[0002] In automated storage and retrieval systems (AS / RS) and intelligent logistics systems, shuttle vehicles, as key handling equipment, directly impact warehousing efficiency and system stability through their path planning. Shuttle vehicles operate on a grid-like track network composed of straight tracks and intersecting nodes, performing inbound / outbound and inter-location transfer tasks. Existing technologies have incorporated ant colony optimization algorithms, which simulate the mechanism of ants releasing pheromones and selecting high-pheromone paths to search for optimal routes among discrete track nodes.

[0003] How can shuttle route planning be improved from a discrete node-time slot space to a continuous spatiotemporal domain while retaining the parallel search and pheromone positive feedback mechanism of the ant colony optimization algorithm? This would eliminate path deviation and conflict prediction distortion caused by time and space discretization, and enable adaptive adjustment of the path under multi-vehicle coupling constraints.

[0004] Existing methods discretize the track network into nodes and time slots of equal length, which leads to an inherent deviation between the actual continuous motion trajectory of the shuttle and the planned path. In narrow alleys and multi-intersection scenarios, this deviation can accumulate into unpredictable hidden conflicts. This invention expresses paths using continuous time curves, constructs a spatiotemporally continuous pheromone field, and introduces joint optimization of random displacement and repulsion potential that decays with iteration. Without relying on repeated tuning of artificial parameters, it enables individual ant colonies to directly search for feasible trajectories with low conflict, short curvature, and high smoothness in the continuous spatiotemporal domain, thereby fundamentally solving the problem of collaborative distortion caused by discrete approximation. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for optimizing the path of a logistics shuttle based on the ant colony optimization algorithm, so as to solve the problems mentioned above.

[0006] The objective of this invention can be achieved through the following technical solutions: The logistics shuttle route optimization method based on ant colony optimization algorithm includes the following steps: S1: Collect the real-time location coordinates of each shuttle in the logistics system, the coordinates of the topology nodes of the track network, the starting and ending positions of the tasks to be executed, and the timestamp of the current moment to obtain the initial spatiotemporal data. S2, based on the initial spatiotemporal data, represents the expected travel path of each shuttle as a continuous time curve, and maps the records of each spatiotemporal point on the continuous time curve that are occupied by the historical path to continuous spatiotemporal pheromone concentration values ​​according to the time decay law. S3 uses continuous time curves as individuals in an artificial ant colony, guided by the spatiotemporal pheromone concentration value, and introduces the repulsion potential based on the predicted trajectories of other shuttles to construct a joint cost including path length, curvature smoothness and repulsion potential. Driven by the joint cost, the control points of each continuous time curve are iteratively adjusted. S4. Sort the spatiotemporal trajectories corresponding to each continuous time curve after iterative adjustment according to their conflict integral values ​​in the spatiotemporal pheromone concentration field, and select the curve with a conflict integral value lower than the threshold and the minimum joint cost as the optimal continuous path of the current shuttle. S5 utilizes the spatiotemporal point information on the optimal continuous path to enhance the spatiotemporal pheromone concentration value at the corresponding location, while converting the optimal continuous path into a continuous speed-time command sequence required for the actual operation of the shuttle, thus completing the path optimization.

[0007] As a further aspect of the present invention: the process of obtaining the spatiotemporal pheromone concentration value is as follows: Using the centerline of each track as a reference, the historical path occupancy points of each shuttle are decomposed into arc length coordinates along the track line and corresponding time coordinates to obtain a spatiotemporal point set; Gaussian diffusion was performed on each point in the spatiotemporal point set along the orbital direction and the time direction using the exponential decay law. Then, all the Gaussian diffused values ​​were superimposed to obtain the initial pheromone concentration at each arc length-time coordinate. The pheromone concentration between adjacent orbital nodes at the same time is interpolated along the orbital line using cubic spline interpolation to form a spatiotemporal pheromone concentration value that is continuous along the orbit and changes continuously with time.

[0008] As a further aspect of the present invention: the Gaussian diffusion specifically includes: Obtain the instantaneous speed of the shuttle at each historical occupancy point, and determine the forward decay step size and backward decay step size along the track direction based on the instantaneous speed, wherein the forward decay step size is larger than the backward decay step size. Using the arc length coordinates of each historical occupied point as the center, the spatial diffusion value at each arc length position is calculated along the forward and backward directions of the track according to the exponential decay law with their respective decay step sizes. At the same time, the temporal diffusion value at each moment is calculated with the time coordinates of the corresponding point as the center according to a fixed time constant. Multiply the spatial diffusion value and the temporal diffusion value at the same spatiotemporal coordinate to obtain the final diffusion value of the corresponding point at the spatiotemporal coordinate. Then, sum the final diffusion values ​​of all historical occupied points at the same spatiotemporal coordinate to complete Gaussian diffusion.

[0009] As a further aspect of the present invention: the iterative adjustment of the control points of each continuous time curve driven by the joint cost specifically includes: Calculate the direction of the partial derivative of the joint cost of the current continuous time curve at each control point with respect to the coordinates of the corresponding point to obtain the descent direction; A random displacement vector that decays with the number of iterations is superimposed on the descent direction to form an exploratory displacement; After moving each control point along the tentative displacement, the joint cost of the continuous time curve is recalculated. If the new joint cost is less than the original joint cost, the move is retained; otherwise, the point is returned to its original position and the magnitude of the random displacement vector is reduced. This process is repeated until multiple iterations fail to reduce the joint cost.

[0010] As a further aspect of the present invention: the provision of the tentative displacement specifically includes: The ratio of the current iteration number to the preset total number of iterations is used as the decay ratio. The initial random amplitude is multiplied by the decay ratio to obtain the current random amplitude. The current random amplitude decreases as the number of iterations increases. For each coordinate direction of each control point, generate a random value between the negative current random amplitude and the positive current random amplitude, and combine the random values ​​of all coordinate directions into a random displacement vector; The descent direction and the random displacement vector are normalized according to their respective magnitudes and then superimposed. The superposition weight of the descent direction is fixed at 1, and the superposition weight of the random displacement vector is equal to the attenuation ratio, thus forming an exploratory displacement.

[0011] As a further aspect of the present invention: selecting the curve with the conflict integral value below the threshold and the minimum joint cost as the optimal continuous path of the current shuttle specifically includes: Each continuous time curve is divided into continuous time segments on the time axis according to a preset duration. The spatiotemporal pheromone concentration value of the time curve is integrated along the orbital line within each time segment to obtain the segment conflict integral of each time segment. The total conflict integral is obtained by summing the conflict integrals of all time segments. Time segments with a total conflict integral higher than the threshold are removed, and the remaining time segments are arranged in ascending order of joint cost. Select the curve with the minimum joint cost from the permutations as the optimal continuous path for the current shuttle.

[0012] As a further aspect of the present invention: obtaining the segment conflict integral for each time segment specifically includes: The projection trajectory of the current continuous time curve within the time segment is used as a reference to divide each arc length into a fixed number of sampling arc length intervals. At the midpoint of each sampling arc length interval, obtain the corresponding orbital position coordinates and the time coordinates of the time curve at the orbital position, and read the spatiotemporal pheromone concentration value at the spatiotemporal coordinates; Multiply the length of each sampling arc length interval by the spatiotemporal pheromone concentration value at the midpoint of the sampling arc length interval, and then add the products of all sampling arc length intervals to obtain the fragment conflict integral within the time segment.

[0013] As a further aspect of the present invention: the method of using spatiotemporal point information on the optimal continuous path to inversely enhance the spatiotemporal pheromone concentration value at the corresponding location specifically includes: The instantaneous velocity at the midpoint of each sampling arc length interval on the optimal continuous path is obtained. Based on the instantaneous velocity, the forward enhancement length and backward enhancement length along the track direction are determined. The forward enhancement length is proportional to the instantaneous velocity, and the backward enhancement length is a fixed minimum value. Taking the spatiotemporal coordinates of the midpoint of each sampling arc length interval as the center, along the orbital direction within the forward enhancement length and backward enhancement length range, the pheromone concentration values ​​at each corresponding spatiotemporal coordinate are increased according to the law of exponential decay with the increase of the arc length from the center. The enhancement contributions corresponding to the midpoints of all sampling arc length intervals on the optimal continuous path are accumulated at the same spatiotemporal coordinates to complete the reverse enhancement.

[0014] A logistics shuttle route optimization system based on ant colony optimization algorithm includes: The data acquisition module collects the real-time location coordinates of each shuttle in the logistics system, the coordinates of the topology nodes of the track network, the start and end positions of the tasks to be executed, and the timestamp of the current moment to obtain the initial spatiotemporal data. The curve mapping module, based on the initial spatiotemporal data, represents the expected travel path of each shuttle as a continuous time curve, and maps the records of each spatiotemporal point on the continuous time curve that are occupied by the historical path to continuous spatiotemporal pheromone concentration values ​​according to the time decay law. The cost iteration module uses continuous time curves as individuals in an artificial ant colony and is guided by the spatiotemporal pheromone concentration value. It also introduces the repulsion potential based on the predicted trajectories of other shuttles to construct a joint cost including path length, curvature smoothness, and repulsion potential. Driven by the joint cost, the control points of each continuous time curve are iteratively adjusted. The optimal selection module sorts the spatiotemporal trajectories corresponding to each continuous time curve after iterative adjustment according to their conflict integral values ​​in the spatiotemporal pheromone concentration field, and selects the curve with a conflict integral value lower than the threshold and the minimum joint cost as the optimal continuous path of the current shuttle. The enhancement conversion module uses the spatiotemporal point information on the optimal continuous path to enhance the spatiotemporal pheromone concentration value at the corresponding location. At the same time, it converts the optimal continuous path into a continuous speed-time command sequence required for the actual operation of the shuttle, thus completing the path optimization.

[0015] The beneficial effects of this invention are: (1) By expressing the shuttle path as a continuous time curve and establishing a spatiotemporally continuous pheromone concentration field, the discretization of time and space in the traditional method is replaced, making the path planning result more in line with the continuous motion characteristics of the actual operation of the shuttle. This effectively reduces the path deviation and hidden conflict caused by discrete approximation in dense track networks, and reduces the probability of collision or deadlock during multi-vehicle cooperative operation.

[0016] (2) By combining the repulsion potential in the joint cost with the random displacement decay mechanism based on the number of iterations, the path optimization process has sufficient global exploration capability in the early stage and can stably converge to the low-cost path in the later stage. At the same time, the optimal path is used to enhance the pheromone concentration in reverse, realizing the adaptive accumulation of historical effective path experience. There is no need to manually adjust the algorithm parameters repeatedly, which reduces the complexity of on-site deployment and maintenance. Attached Figure Description

[0017] The invention will now be further described with reference to the accompanying drawings.

[0018] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a flowchart of the process for obtaining the spatiotemporal pheromone concentration value in this invention; Figure 3 This is a system block diagram of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Please see Figure 1 As shown, this invention is a logistics shuttle route optimization method based on ant colony optimization algorithm, including the following steps: S1: Collect the real-time location coordinates of each shuttle in the logistics system, the coordinates of the topology nodes of the track network, the starting and ending positions of the tasks to be executed, and the timestamp of the current moment to obtain the initial spatiotemporal data. S2, based on the initial spatiotemporal data, represents the expected travel path of each shuttle as a continuous time curve, and maps the records of each spatiotemporal point on the time curve occupied by the historical path to continuous spatiotemporal pheromone concentration values ​​according to the time decay law. S3 uses continuous time curves as individuals in an artificial ant colony, guided by the spatiotemporal pheromone concentration value, and introduces the repulsion potential based on the predicted trajectories of other shuttles to construct a joint cost including path length, curvature smoothness and repulsion potential. Driven by the joint cost, the control points of each continuous time curve are iteratively adjusted. S4. Sort the spatiotemporal trajectories corresponding to each continuous time curve after iterative adjustment according to their conflict integral values ​​in the spatiotemporal pheromone concentration field, and select the curve with a conflict integral value lower than the threshold and the minimum joint cost as the optimal continuous path of the current shuttle. S5 utilizes the spatiotemporal point information on the optimal continuous path to enhance the spatiotemporal pheromone concentration value at the corresponding location, while converting the optimal continuous path into a continuous speed-time command sequence required for the actual operation of the shuttle, thus completing the path optimization.

[0021] In S1, the real-time location coordinates of each shuttle in the logistics system, the coordinates of the track network topology nodes, the start and end positions of the tasks to be executed, and the current timestamp are collected to obtain initial spatiotemporal data, specifically including: At the logistics shuttle operation site, each shuttle is equipped with an incremental rotary encoder and an RFID reader. The rotary encoder is coaxially connected to the shuttle's drive wheel axle to measure the number and direction of wheel rotations in real time; the RFID reader scans RFID tags embedded at track nodes. The track network consists of multiple straight tracks and cross-shaped tracks, with uniquely numbered RFID tags pre-embedded at the intersections and endpoints of each track. When a shuttle passes a RFID tag, the reader reads the tag number and records the current time. Combined with the number of pulses accumulated by the encoder since the last tag reading, the real-time position coordinates of the shuttle in the track coordinate system are calculated. These coordinates include the track line markings and the arc length distance from the starting point of that track.

[0022] The coordinates of the track network topology nodes are pre-stored in a configuration file on the host computer. This topology data includes a unique identifier for each node, the track line number to which the node belongs, the node's two-dimensional projected coordinates in the track plane, and the connection relationships between nodes. Nodes are categorized as ordinary waypoints, intersections, charging stops, and in / out stations. The start and end points of the tasks to be executed also come from the task list issued by the host computer. Each task specifies that a shuttle vehicle needs to pick up goods from the starting point and transport them to the ending point, with both the start and end points corresponding to the node coordinates in the track network topology.

[0023] The current timestamp is provided by the system clock of the industrial control computer, which is synchronized with the clock of the programmable logic controller (PLC) via a real-time Ethernet protocol, achieving a synchronization accuracy of within 1 millisecond. Each time data acquisition is triggered, the shuttle's wireless communication module packages the real-time position coordinates, the current timestamp, and the RFID tag number of the most recently passed location, and sends it to the host computer according to the User Datagram Protocol (UDP). Simultaneously, the host computer reads the start and end positions of all unfinished tasks performed by the shuttle from the task queue.

[0024] The host computer stores the received real-time location coordinates of the shuttle, the coordinates of the track network topology nodes, the start and end points of the tasks to be executed, and the current timestamp in a unified data structure. Specifically, using the shuttle number and timestamp as a joint primary key, each record contains the arc length coordinates at that moment, the track line number, the coordinates of the nearest node, and the node coordinates of the next task start and end points assigned to that shuttle. The integrated data structure constitutes the initial spatiotemporal data, which is used for subsequent path planning.

[0025] Please see Figure 2 As shown, in S2, based on the initial spatiotemporal data, the expected travel path of each shuttle is represented as a continuous time curve. The records of historical paths occupying each spatiotemporal point on the time curve are mapped to continuous spatiotemporal pheromone concentration values ​​according to the time decay law, specifically including: Based on the initial spatiotemporal data, the process of representing the expected travel path of each shuttle as a continuous-time curve is as follows: Based on the track network topology stored in the host computer, a one-dimensional arc-length coordinate system is established for each track line with its midline as the reference, and the zero point of the arc length is set at the endpoint of the track line. For each shuttle, the starting and ending arc-length coordinates of its task to be performed are combined with the expected departure time from the starting point and the expected arrival time at the ending point, and a curve is fitted using a cubic polynomial with time as the independent variable and arc length as the dependent variable. This curve is the continuous-time curve. This curve is continuously differentiable within its defined time interval, and its first derivative is the instantaneous velocity of the shuttle traveling along the track, while the second derivative corresponds to acceleration.

[0026] The process of obtaining the spatiotemporal pheromone concentration value is as follows: Using the centerline of each track as a reference, the historical path occupancy points of each shuttle recorded in the host computer are decomposed into arc length coordinates along the track line and corresponding time coordinates to form a spatiotemporal point set. Each point contains three values: arc length value, time value, and instantaneous speed value of the shuttle at that point.

[0027] Gaussian diffusion is applied to each point in the spatiotemporal point set along the orbital direction and the time direction using an exponential decay law. Specifically, for each historically occupied point, the spatial influence is decayed exponentially from its arc length coordinate to both sides of the orbit, and the temporal influence is decayed exponentially from its time coordinate to both sides of time. The final contribution value of the point to the surrounding spatiotemporal domain is obtained by multiplying the spatial diffusion value and the temporal diffusion value.

[0028] The contribution values ​​of all historical occupancy points are superimposed at the same spatiotemporal coordinate to obtain the initial pheromone concentration at each arc length-time coordinate. Finally, the pheromone concentration between adjacent orbital nodes at the same moment is interpolated along the orbital line using cubic spline interpolation, so that there is a corresponding pheromone concentration value at any arc length coordinate. At the same arc length coordinate, the pheromone concentration at different times also changes continuously through linear interpolation on the time axis, thus forming a spatiotemporal pheromone concentration value that is continuous along the orbit and changes continuously with time.

[0029] The specific implementation of the Gaussian diffusion described above is as follows: The instantaneous velocity value of the shuttle at each historical occupancy point is obtained, and the forward decay step size and backward decay step size along the track direction are determined based on this instantaneous velocity. The forward decay step size is equal to the instantaneous velocity value multiplied by a first fixed coefficient, which is set to 1.5. The backward decay step size is equal to the instantaneous velocity value multiplied by a second fixed coefficient, which is set to 0.8, such that the forward decay step size is greater than the backward decay step size.

[0030] Centered on the arc length coordinates of each historical occupancy point, the spatial diffusion value is calculated along the forward and backward directions of the track according to its respective decay step size: For any point in the forward direction of the track, its spatial diffusion value is equal to the negative first power of the natural constant multiplied by the arc length distance from that point to the center point divided by the forward decay step size; for any point in the backward direction of the track, its spatial diffusion value is equal to the negative first power of the natural constant multiplied by the arc length distance from that point to the center point divided by the backward decay step size. Centered on the time coordinates of that point, the time diffusion value at each moment is calculated according to a fixed time constant, which is 0.5 seconds. The time diffusion value is equal to the negative first power of the natural constant multiplied by the absolute value of the time difference divided by the fixed time constant.

[0031] Multiplying the spatial diffusion value and temporal diffusion value at the same spatiotemporal coordinate yields the final diffusion value of that historical occupancy point at that coordinate. For each spatiotemporal coordinate to be calculated, i.e., a specific arc length and time value, the final diffusion values ​​of all historical occupancy points at that coordinate need to be accumulated. The accumulated result is the initial pheromone concentration at that spatiotemporal coordinate. The above calculation process is executed offline on the host computer, and the historical path occupancy record is updated once before each new task begins.

[0032] After calculating the initial pheromone concentration, interpolation is required to obtain the pheromone concentration values ​​at any arc length position and any time along the track. Specifically, for the same time, the pheromone concentration values ​​at two adjacent known arc length nodes on the track at that time are selected, and a cubic spline function is used for interpolation between these two nodes, ensuring that the first and second derivatives of the interpolated function are continuous at the nodes. For the same arc length position between different times, a linear interpolation method is used, that is, the concentration value at the intermediate time is calculated proportionally based on the pheromone concentration values ​​of the two consecutive times and the time difference. After the above two-dimensional interpolation, a continuous spatiotemporal pheromone concentration field is obtained. Each value in this field is uniquely determined by the arc length coordinate and the time coordinate, representing the relative probability that a given track position at a given time is occupied by a historical path. This spatiotemporal pheromone concentration field will serve as guiding information for subsequent path searching by individual artificial ant colonies.

[0033] In S3, continuous-time curves are used as individuals in the artificial ant colony, guided by spatiotemporal pheromone concentration values. Simultaneously, a repulsion potential based on predicted trajectories from other shuttles is introduced to construct a joint cost including path length, curvature smoothness, and repulsion potential. Driven by this joint cost, the control points of each continuous-time curve are iteratively adjusted, specifically including: During the iterative adjustment of the continuous-time curve, the partial derivative direction of the joint cost at each control point with respect to that point's position coordinates is calculated. The joint cost consists of a weighted sum of path length cost, curvature smoothness cost, and repulsion potential cost. For any control point, its position coordinates are slightly perturbed along the track arc length direction. The change in joint cost before and after the perturbation is calculated, and the change is divided by the perturbation step size to obtain the approximate partial derivative value in that direction. Specifically, assuming a perturbation step size of 0.001 meters, the arc length coordinates of the control point are perturbed forward and backward, and the increment of the joint cost is recorded. The forward increment minus the backward increment divided by twice the perturbation step size gives the partial derivative value of the joint cost with respect to the arc length coordinates at that control point. The partial derivative values ​​of all control points are combined into a vector, the direction of which is the descent direction, denoted as the descent direction vector. Each component of this vector corresponds to the partial derivative value of a control point in a coordinate direction.

[0034] In addition to the descent direction, a random displacement vector that decays with the number of iterations needs to be superimposed to enhance the global search capability. The random displacement vector is constructed as follows: Let the current iteration number be the... The preset total number of iterations is [number]. ,Pick The value is 200. Attenuation ratio. Calculate according to the following formula: ; in, This represents the current number of iterations completed. The preset total number of iterations, The value range gradually increases from near 0 to 1. The initial random amplitude is set to 0.05 meters. Multiplying the initial random amplitude by the attenuation ratio yields the current random amplitude; that is, the current random amplitude equals the initial random amplitude multiplied by the attenuation ratio. For each control point, the range of its position coordinates along the track arc length is finite. In this implementation, each control point has only one coordinate direction, namely the arc length coordinate. A uniformly distributed random value between the negative and positive current random amplitudes is generated; this value is the random displacement component of the control point. The random displacement components of all control points are combined into a random displacement vector.

[0035] The descent direction vector and the random displacement vector are normalized separately, that is, each vector is divided by its own magnitude to obtain the unit descent direction vector and the unit random displacement vector. Then, the tentative displacement vector is synthesized as follows: ; in, For tentative displacement vectors, Let α be the unit descent direction vector, and α be the attenuation ratio. The random displacement vector is a unit displacement vector. The superposition weight in the descent direction is fixed at 1, and the superposition weight of the random displacement vector is equal to the attenuation ratio α. Each component of the tentative displacement vector is added to the current arc length coordinate of the corresponding control point to obtain the new control point position after the movement.

[0036] For each control point, after calculating the trial displacement using the method described above, all control points are simultaneously moved to the new position, forming a new continuous-time curve. The joint cost of this new curve is recalculated. If the new joint cost is less than the original joint cost, the move is retained, and the new curve is used as the current curve in the next iteration; if the new joint cost is greater than or equal to the original joint cost, all control points are returned to their original positions, and the current random amplitude is reduced to half of its original value (i.e., multiplied by 0.5). Then, random values ​​are regenerated, and trial displacements are constructed again for further attempts.

[0037] Repeat the above iterative process until the joint cost cannot be reduced after 20 consecutive iterations. At this point, stop the iteration and output the current continuous-time curve as the final path of the individual ant. During this iterative process, the descent direction is recalculated with each iteration, and the random displacement vector is regenerated each time a move is attempted. Attenuation ratio. With iteration number As the gradient increases, the algorithm gains a greater capacity for random exploration in the early stages and gradually converges to the gradient descent direction in the later stages.

[0038] In S4, the spatiotemporal trajectories corresponding to each continuous time curve after iterative adjustment are sorted according to their conflict integral values ​​in the spatiotemporal pheromone concentration field. The curve with a conflict integral value below a threshold and the minimum joint cost is selected as the optimal continuous path for the current shuttle. Specifically, this includes: After completing the iterative adjustment of each continuous time curve, it is necessary to evaluate the degree of conflict between each curve and the historical path occupancy, that is, to calculate the conflict integral value of the curve in the spatiotemporal pheromone concentration field.

[0039] The specific method is as follows: The total time range covered by each continuous time curve, from the start time to the end time, is divided into continuous time segments of a preset duration of 0.5 seconds. If the length of the last time segment is less than 0.5 seconds, it is processed according to its actual length. The preset duration of 0.5 seconds is determined based on the average time for the shuttle to pass through a track intersection node. This value ensures that the kinematic characteristics of the curve are approximately constant within each time segment, while avoiding the smoothing out of local conflict peaks due to excessively long time segments.

[0040] For each time segment, the segment conflict integral is obtained by integrating along the projection trajectory of the continuous time curve on the track within the corresponding time interval. The integration process is as follows: Determine the arc length range covered by the continuous time curve within the time segment, from the initial arc length to the final arc length. Using the arc length distance between adjacent control points as a reference, each arc length is divided into a fixed number of sampling arc length intervals, with the fixed number being 10. For example, if the arc length distance between two adjacent control points is 1.2 meters, it is divided into 10 intervals, each with a length of 0.12 meters. At the midpoint of each sampling arc length interval, obtain the track position coordinates corresponding to that point and the time coordinates of the continuous time curve at that position. Read the pheromone concentration value corresponding to that spatiotemporal coordinate from the constructed spatiotemporal pheromone concentration field. Multiply the length of each sampling arc length interval by the pheromone concentration value at its midpoint to obtain a product term. Then, add the product terms of all sampling arc length intervals together. The result is the segment conflict integral within the time segment.

[0041] After calculating the segment conflict integrals of the continuous-time curve across all time segments using the method described above, the segment conflict integrals of all time segments are summed to obtain the total conflict integral of the curve. The total conflict integral characterizes the overall spatiotemporal overlap between the curve and historical path occupancy. A conflict integral threshold is pre-set, and this threshold is determined as follows: the total conflict integrals of the most recent 100 successfully executed task paths are statistically analyzed, their average value is calculated, and 1.5 times this average value is used as the conflict integral threshold for the current batch. If the total conflict integral of a curve is higher than this threshold, it indicates a high degree of spatiotemporal conflict between the curve and historical path occupancy, which is likely to cause collisions or deadlocks in actual operation, and therefore it is discarded.

[0042] After removing curves with total conflict integrals exceeding a threshold, the remaining continuous-time curves are arranged in ascending order of their joint costs. The joint cost combines path length cost, curvature smoothness cost, and repulsion potential cost. The curve with the lowest joint cost in the arrangement is the optimal continuous path for the current shuttle. If the number of remaining curves after removal is zero, the threshold is appropriately relaxed to 1.2 times the original threshold, and the removal and sorting operations are repeated until at least one curve is available.

[0043] The process of selecting the optimal continuous path described above is performed independently for each shuttle. When multiple shuttles participate in the task simultaneously, the optimal continuous path for each shuttle must be selected sequentially according to the sorting rules in step four. That is, the optimal path of the shuttle with the lowest joint cost is selected first, and then the other shuttles are processed in turn to avoid mutual interference when multiple shuttles select routes. After the selection is completed, each optimal continuous path is output.

[0044] In S5, the spatiotemporal point information on the optimal continuous path is used to amplify the spatiotemporal pheromone concentration at the corresponding location. Simultaneously, the optimal continuous path is converted into a continuous speed-time command sequence required for the actual operation of the shuttle, thus completing path optimization. Specifically, this includes: After obtaining the optimal continuous path, the instantaneous velocity at the midpoint of each sampling arc length interval along the path is first extracted. The sampling arc length interval is divided in the same way as when calculating the segment conflict integral, that is, the arc length between each adjacent control point is divided into 10 intervals, and the length of each interval is one-tenth of the arc length distance between adjacent control points. The instantaneous velocity at the midpoint of each interval is obtained by taking the first derivative of the optimal continuous time curve, which has been calculated during curve fitting, and represents the rate of change of the arc length at that point with respect to time. Substituting the arc length coordinates of each midpoint into the velocity function, the corresponding instantaneous velocity value is output in meters per second.

[0045] The forward and backward reinforcement lengths along the track direction are determined based on the instantaneous velocity at each midpoint. The forward reinforcement length equals the instantaneous velocity multiplied by a fixed time factor of 0.3 seconds; for example, when the instantaneous velocity is 1.5 meters per second, the forward reinforcement length is 0.45 meters. The backward reinforcement length takes a fixed minimum value of 0.2 meters and is independent of the instantaneous velocity. The fixed time factor of 0.3 seconds is determined based on the response delay time of the shuttle control system, and the fixed minimum value of 0.2 meters corresponds to half the length of the shuttle body. The forward reinforcement length is always greater than or equal to the backward reinforcement length.

[0046] Using the spatiotemporal coordinates of the midpoint of each sampling arc length interval as the center, pheromone concentration enhancement is performed along the orbital direction within the forward and backward enhancement lengths. Specifically, for each midpoint, let its arc length coordinate be L0 and its time coordinate be T0. Within the arc length interval from (L0 minus the backward enhancement length) to (L0 plus the forward enhancement length), an enhancement point is taken every 0.05 meters. For each enhancement point, the absolute value of its arc length distance to the center point is calculated, and this absolute value is divided by the enhancement attenuation constant 0.15 meters to obtain the ratio. Then, the negative power of the natural constant is taken to obtain the spatial attenuation coefficient. Simultaneously, in the time direction, with T0 as the center, 0.2 seconds are taken forward and backward, and a time point is taken every 0.02 seconds. The absolute value of the time difference is calculated and divided by the time attenuation constant 0.1 seconds to obtain the time ratio. Similarly, the negative power of the natural constant is taken to obtain the time attenuation coefficient. Multiply the spatial attenuation coefficient by the temporal attenuation coefficient, and then multiply by a fixed enhancement amplitude of 0.01 to obtain the increase in pheromone concentration at that spatiotemporal coordinate.

[0047] The calculated increase is then added to the original value of the spatiotemporal pheromone concentration field. For the enhancement contribution corresponding to the midpoint of all sampling arc-length intervals on the optimal continuous path, the summation is performed at the same spatiotemporal coordinate. If multiple midpoints enhance the same spatiotemporal coordinate, all increases are summed and added to the original concentration value. After processing all midpoints, a reverse enhancement of the path pheromone concentration is achieved, increasing the probability of this optimal continuous path being selected in future iterations.

[0048] The optimal continuous path is converted into a continuous speed-time command sequence required for the actual operation of the shuttle. The conversion process is as follows: The time axis of the optimal continuous time curve is discretized from the start time to the end time with a fixed period of 0.02 seconds, resulting in a series of discrete moments. For each discrete moment, the corresponding arc length position and instantaneous speed are calculated using a curve function. These discrete moments and their corresponding instantaneous speeds are paired to form speed-time point pairs. These point pairs are then arranged in chronological order, with linear interpolation used between adjacent point pairs to generate a smooth speed change curve. Finally, the arc length position and desired speed value for each discrete moment are sent sequentially to the shuttle's controller, which adjusts the motor speed accordingly, causing the shuttle to travel along the optimal continuous path. This completes the path optimization.

[0049] Please see Figure 3 As shown, the logistics shuttle route optimization system based on ant colony optimization algorithm includes: The data acquisition module collects the real-time location coordinates of each shuttle in the logistics system, the coordinates of the topology nodes of the track network, the start and end positions of the tasks to be executed, and the timestamp of the current moment to obtain the initial spatiotemporal data. The curve mapping module, based on the initial spatiotemporal data, represents the expected travel path of each shuttle as a continuous time curve, and maps the records of historical paths occupying each spatiotemporal point on the time curve into continuous spatiotemporal pheromone concentration values ​​according to the time decay law. The cost iteration module uses continuous time curves as individuals in an artificial ant colony and is guided by the spatiotemporal pheromone concentration value. It also introduces the repulsion potential based on the predicted trajectories of other shuttles to construct a joint cost including path length, curvature smoothness, and repulsion potential. Driven by the joint cost, the control points of each continuous time curve are iteratively adjusted. The optimal selection module sorts the spatiotemporal trajectories corresponding to each continuous time curve after iterative adjustment according to their conflict integral values ​​in the spatiotemporal pheromone concentration field, and selects the curve with a conflict integral value lower than the threshold and the minimum joint cost as the optimal continuous path of the current shuttle. The enhancement conversion module uses the spatiotemporal point information on the optimal continuous path to enhance the spatiotemporal pheromone concentration value at the corresponding location. At the same time, it converts the optimal continuous path into a continuous speed-time command sequence required for the actual operation of the shuttle, thus completing the path optimization.

[0050] The working principle of this invention is as follows: Real-time location coordinates of the shuttle, coordinates of the track network topology nodes, the start and end points of the task, and timestamps are collected to obtain initial spatiotemporal data. Based on this data, the expected travel path of each shuttle is represented as a continuous time curve, and the records of historical paths occupying each spatiotemporal point on the curve are mapped to continuous spatiotemporal pheromone concentration values ​​according to the time decay law. Each continuous time curve is treated as an individual in an artificial ant colony, guided by the spatiotemporal pheromone concentration value. Simultaneously, a repulsion potential based on the predicted trajectories of other shuttles is introduced to construct a joint cost including path length, curvature smoothness, and repulsion potential. Under the drive of this joint cost, the control points of each curve are iteratively adjusted. The spatiotemporal trajectories corresponding to each adjusted curve are sorted according to their conflict integral values ​​in the spatiotemporal pheromone concentration field. The curve with a conflict integral value below a threshold and the smallest joint cost is selected as the optimal continuous path for the current shuttle. The spatiotemporal point information on this optimal path is used to amplify the spatiotemporal pheromone concentration value at the corresponding location. Simultaneously, the optimal continuous path is converted into a continuous speed-time command sequence required for the actual operation of the shuttle, completing path optimization.

[0051] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A method for optimizing the path of a logistics shuttle based on ant colony optimization algorithm, characterized in that, Includes the following steps: S1: Collect the real-time location coordinates of each shuttle in the logistics system, the coordinates of the topology nodes of the track network, the starting and ending positions of the tasks to be executed, and the timestamp of the current moment to obtain the initial spatiotemporal data. S2, based on the initial spatiotemporal data, represents the expected travel path of each shuttle as a continuous time curve, and maps the records of each spatiotemporal point on the continuous time curve that are occupied by the historical path to continuous spatiotemporal pheromone concentration values ​​according to the time decay law. S3 uses continuous time curves as individuals in an artificial ant colony, guided by the spatiotemporal pheromone concentration value, and introduces the repulsion potential based on the predicted trajectories of other shuttles to construct a joint cost including path length, curvature smoothness and repulsion potential. Driven by the joint cost, the control points of each continuous time curve are iteratively adjusted. S4. Sort the spatiotemporal trajectories corresponding to each continuous time curve after iterative adjustment according to their conflict integral values ​​in the spatiotemporal pheromone concentration field, and select the curve with a conflict integral value lower than the threshold and the minimum joint cost as the optimal continuous path of the current shuttle. S5 utilizes the spatiotemporal point information on the optimal continuous path to enhance the spatiotemporal pheromone concentration value at the corresponding location, while converting the optimal continuous path into a continuous speed-time command sequence required for the actual operation of the shuttle, thus completing the path optimization.

2. The logistics shuttle route optimization method based on ant colony optimization algorithm according to claim 1, characterized in that, The process for obtaining the spatiotemporal pheromone concentration value is as follows: Using the centerline of each track as a reference, the historical path occupancy points of each shuttle are decomposed into arc length coordinates along the track line and corresponding time coordinates to obtain a spatiotemporal point set; Gaussian diffusion was performed on each point in the spatiotemporal point set along the orbital direction and the time direction using the exponential decay law. Then, all the Gaussian diffused values ​​were superimposed to obtain the initial pheromone concentration at each arc length-time coordinate. The pheromone concentration between adjacent orbital nodes at the same time is interpolated along the orbital line using cubic spline interpolation to form a spatiotemporal pheromone concentration value that is continuous along the orbit and changes continuously with time.

3. The logistics shuttle route optimization method based on ant colony optimization algorithm according to claim 2, characterized in that, The Gaussian diffusion process specifically includes: Obtain the instantaneous speed of the shuttle at each historical occupancy point, and determine the forward decay step size and backward decay step size along the track direction based on the instantaneous speed, wherein the forward decay step size is larger than the backward decay step size. Using the arc length coordinates of each historical occupied point as the center, the spatial diffusion value at each arc length position is calculated along the forward and backward directions of the track according to the exponential decay law with their respective decay step sizes. At the same time, the temporal diffusion value at each moment is calculated with the time coordinates of the corresponding point as the center according to a fixed time constant. Multiply the spatial diffusion value and the temporal diffusion value at the same spatiotemporal coordinate to obtain the final diffusion value of the corresponding point at the spatiotemporal coordinate. Then, sum the final diffusion values ​​of all historical occupied points at the same spatiotemporal coordinate to complete Gaussian diffusion.

4. The logistics shuttle route optimization method based on ant colony optimization algorithm according to claim 1, characterized in that, The iterative adjustment of control points for each continuous-time curve driven by joint cost specifically includes: Calculate the direction of the partial derivative of the joint cost of the current continuous time curve at each control point with respect to the coordinates of the corresponding point to obtain the descent direction; A random displacement vector that decays with the number of iterations is superimposed on the descent direction to form an exploratory displacement; After moving each control point along the tentative displacement, the joint cost of the continuous time curve is recalculated. If the new joint cost is less than the original joint cost, the move is retained; otherwise, the point is returned to its original position and the magnitude of the random displacement vector is reduced. This process is repeated until multiple iterations fail to reduce the joint cost.

5. The logistics shuttle route optimization method based on ant colony optimization algorithm according to claim 4, characterized in that, The tentative displacement specifically includes: The ratio of the current iteration number to the preset total number of iterations is used as the decay ratio. The initial random amplitude is multiplied by the decay ratio to obtain the current random amplitude. The current random amplitude decreases as the number of iterations increases. For each coordinate direction of each control point, generate a random value between the negative current random amplitude and the positive current random amplitude, and combine the random values ​​of all coordinate directions into a random displacement vector; The descent direction and the random displacement vector are normalized according to their respective magnitudes and then superimposed. The superposition weight of the descent direction is fixed at 1, and the superposition weight of the random displacement vector is equal to the attenuation ratio, thus forming an exploratory displacement.

6. The logistics shuttle route optimization method based on ant colony optimization algorithm according to claim 1, characterized in that, The selection of the curve with the conflict integral value below the threshold and the minimum joint cost as the optimal continuous path for the current shuttle specifically includes: Each continuous time curve is divided into continuous time segments on the time axis according to a preset duration. The spatiotemporal pheromone concentration value of the time curve is integrated along the orbital line within each time segment to obtain the segment conflict integral of each time segment. The total conflict integral is obtained by summing the conflict integrals of all time segments. Time segments with a total conflict integral higher than the threshold are removed, and the remaining time segments are arranged in ascending order of joint cost. Select the curve with the minimum joint cost from the permutations as the optimal continuous path for the current shuttle.

7. The logistics shuttle route optimization method based on ant colony optimization algorithm according to claim 6, characterized in that, The process of obtaining the segment conflict integral for each time segment specifically includes: The projection trajectory of the current continuous time curve within the time segment is used as a reference to divide each arc length into a fixed number of sampling arc length intervals. At the midpoint of each sampling arc length interval, obtain the corresponding orbital position coordinates and the time coordinates of the time curve at the orbital position, and read the spatiotemporal pheromone concentration value at the spatiotemporal coordinates; Multiply the length of each sampling arc length interval by the spatiotemporal pheromone concentration value at the midpoint of the sampling arc length interval, and then add the products of all sampling arc length intervals to obtain the fragment conflict integral within the time segment.

8. The logistics shuttle route optimization method based on ant colony optimization algorithm according to claim 1, characterized in that, The method of using spatiotemporal point information on the optimal continuous path to inversely enhance the spatiotemporal pheromone concentration value at the corresponding location specifically includes: The instantaneous velocity at the midpoint of each sampling arc length interval on the optimal continuous path is obtained. Based on the instantaneous velocity, the forward enhancement length and backward enhancement length along the track direction are determined. The forward enhancement length is proportional to the instantaneous velocity, and the backward enhancement length is a fixed minimum value. Taking the spatiotemporal coordinates of the midpoint of each sampling arc length interval as the center, along the orbital direction within the forward enhancement length and backward enhancement length range, the pheromone concentration values ​​at each corresponding spatiotemporal coordinate are increased according to the law of exponential decay with the increase of the arc length from the center. The enhancement contributions corresponding to the midpoints of all sampling arc length intervals on the optimal continuous path are accumulated at the same spatiotemporal coordinates to complete the reverse enhancement.

9. A logistics shuttle route optimization system based on ant colony optimization algorithm, characterized in that, The method for optimizing the path of a logistics shuttle based on the ant colony optimization algorithm as described in any one of claims 1-8 includes: The data acquisition module collects the real-time location coordinates of each shuttle in the logistics system, the coordinates of the topology nodes of the track network, the start and end positions of the tasks to be executed, and the timestamp of the current moment to obtain the initial spatiotemporal data. The curve mapping module, based on the initial spatiotemporal data, represents the expected travel path of each shuttle as a continuous time curve, and maps the records of each spatiotemporal point on the continuous time curve that are occupied by the historical path to continuous spatiotemporal pheromone concentration values ​​according to the time decay law. The cost iteration module uses continuous time curves as individuals in an artificial ant colony and is guided by the spatiotemporal pheromone concentration value. It also introduces the repulsion potential based on the predicted trajectories of other shuttles to construct a joint cost including path length, curvature smoothness, and repulsion potential. Driven by the joint cost, the control points of each continuous time curve are iteratively adjusted. The optimal selection module sorts the spatiotemporal trajectories corresponding to each continuous time curve after iterative adjustment according to their conflict integral values ​​in the spatiotemporal pheromone concentration field, and selects the curve with a conflict integral value lower than the threshold and the minimum joint cost as the optimal continuous path of the current shuttle. The enhancement conversion module uses the spatiotemporal point information on the optimal continuous path to enhance the spatiotemporal pheromone concentration value at the corresponding location. At the same time, it converts the optimal continuous path into a continuous speed-time command sequence required for the actual operation of the shuttle, thus completing the path optimization.

Images