A global path update forward access reference continuation method and system for a distributed weak tracking mobile platform

Abstract

The application relates to a global path update forward access reference splicing method and system for a distributed weak tracking mobile platform, which comprises the following steps: receiving a path update package and freezing a current pose, a speed, a reference path, a task generation and an interface state; performing event classification on the path update package and forming a state modification authority mapping; taking the current pose as an access reference and calculating a selectable advance radius; calculating a dynamic access boundary according to a weak tracking deviation and the advance radius; selecting an exit anchor point or entering an anchor point according to an event classification result; outputting a directly spliced reference path, a transition reference path or a degraded processing event according to anchor point continuity grading; outputting a complete interface field to a local planning end and submitting a state or keeping an old reference according to an acceptance or rejection result. The application can guarantee the forwardness and continuity of reference access, prevent path progress from being rolled back and is suitable for a distributed weak tracking mobile platform navigation system.

Description

Technical Field

[0001] This invention belongs to the technical fields of autonomous navigation for mobile robots, distributed global path update processing, local planning preprocessing, real-time global reference path, path access control for weak tracking platforms, and trajectory planning interface. It relates to a method and system for forward access reference continuation of global path update for distributed weak tracking mobile platforms. Background Technology

[0002] Mobile robot navigation systems typically consist of a global planning unit, a local planning unit, and an execution unit. In a distributed deployment scenario, the global planning unit can run on a first computing unit or a remote computing unit, while the local planning unit can run on a second computing unit or a real-time computing unit located near the sensors and actuators, such as... Figure 1 As shown.

[0003] Since the global planning end and the local planning end belong to different computing units, the global path update may go through path calculation, message encapsulation, network transmission, relay forwarding, queue waiting and receiving end scheduling from generation to receipt by the local planning end. Therefore, there is a path update delay and arrival time jitter that cannot be ignored.

[0004] For wheeled robots, wheeled-legged hybrid platforms, or mobile platforms with lag in steering response, there may be slight lateral deviations, heading lags, or start-up response lags between the actual execution trajectory and the global reference path. If the local planning side directly connects to the first point of the new path that arrives late, it may cause the reference point to lag behind the robot; if the current reference is cleared as soon as the path is reached, it may cause abrupt changes in the local planning input.

[0005] Therefore, there is an urgent need for a forward access reference continuation method suitable for distributed global path update scenarios to solve the current problems. Summary of the Invention

[0006] The purpose of this invention is to address the shortcomings of existing technologies by providing a global path update forward access reference continuation method and system for distributed weak tracking mobile platforms, thereby resolving the aforementioned problems in existing technologies.

[0007] The above-mentioned technical objectives of the present invention will be achieved through the technical solutions described below.

[0008] A global path update forward access reference continuation method for distributed weak tracking mobile platforms includes the following steps: S1. Receive the global path update packet and freeze the current pose, current velocity, current reference path, task generation, and local planning interface state; S2. Classify the events of the path update package according to the path validity, target change distance, path summary and task generation, and determine the status modification permissions corresponding to each event type, forming a mapping from path update events to reference path status modification permissions; S3. Use the current pose as the access reference, and calculate the optional advance radius based on the delivery delay and the current speed when there is an observable delivery delay; S4. Determine the weak tracking deviation based on the minimum distance from the current pose to the unexecuted segment of the current reference path, and calculate the dynamic access boundary based on the weak tracking deviation and the optional advance radius; S5. Select either exit anchor point or enter anchor point based on the event classification results; S6. Based on the distance between the exit anchor point and the entry anchor point, the heading difference, and the obstacle risk, output a direct splicing reference path, a transitional reference path, or a downgraded processing event; S7. Output the reference path, event type, task generation, forward anchor index, spatial anchor position, exit anchor, entry anchor, dynamic access boundary, optional advance radius, degradation flag and path failure flag to the local planning end, and submit the current reference status or keep the old reference based on the acceptance or rejection result of the local planning end.

[0009] In addition to the aspects described above and any possible implementation, a further implementation is provided in which the freezing in S1 also includes the submitted anchor point index and the most recent acceptance or rejection status of the local planning end; the freezing uses the same set of current pose, current velocity and current reference path within a processing cycle.

[0010] In addition to the aspects described above and any possible implementation, a further implementation is provided, wherein the event classification in S2 includes keep-alive events, current task rolling update events, new task switching events, path failure events, pending confirmation events, and degradation processing events; the process of forming a mapping from path update events to reference path status modification permissions is specifically as follows: keep-alive events allow modification of availability timestamps; rolling update events allow modification of unexecuted preceding reference paths and anchor fields; new task switching events allow modification of task generations and new reference entry points; path failure events allow modification of failure flags; pending confirmation events do not submit new references; and degradation processing events allow modification of degradation flags and replanning requests.

[0011] In addition to the aspects and any possible implementations described above, a further implementation is provided in which the optional advance radius in step 3 is determined as follows: the difference between the reference path sending time and the local planning end receiving time is recorded N times, the average value is taken to obtain the reference delivery delay, the reference delivery delay is subjected to amplitude limiting processing and multiplied by the modulus of the current speed to obtain the optional advance radius; if the delivery delay is unmeasurable, the jitter exceeds a preset threshold or the current speed is unavailable, the optional advance radius is zero, where N is a natural number.

[0012] In addition to the aspects described above and any possible implementation, a further implementation is provided in which the dynamic access boundary in S4 is determined according to the following relationship: based on the basic access radius, the product of weak tracking deviation and deviation weight and the product of optional advance radius and advance radius weight are superimposed, and the calculation result is cropped to between the preset minimum access boundary and the maximum access boundary; wherein the weak tracking deviation is the minimum distance from the current pose to the unexecuted segment of the current reference path.

[0013] As described above, and in accordance with any possible implementation, an implementation is further provided, wherein S5 specifically includes: S51. When the event classification result is the current task is being updated, construct a candidate set on the new path that satisfies the forward constraint and whose distance does not exceed the sum of the dynamic access boundary and the optional advance radius. Select the point with the longest arc length progress from the candidate set as the forward anchor point, and output the forward anchor point index and the spatial anchor point position at the same time. S52. When the event classification result is a new task switch, select the exit anchor point in the unexecuted segment before the current reference path, and select the entry anchor point in the accessible segment before the new path.

[0014] In addition to the aspects described above and any possible implementation, a further implementation is provided in which the exit anchor point in S52 is selected from the unexecuted segments preceding the current reference path; and the entry anchor point is selected from the accessible segments preceding the new path.

[0015] In addition to the aspects described above and any possible implementation, a further implementation is provided in which the reference path in S7 is the sum of the intermediate transition reference path and the direct splicing reference path.

[0016] In addition to the aspects described above and any possible implementation, a further implementation is provided in which the method for constructing the transition reference path in S6 is as follows: a transition segment is inserted between the exit anchor point and the entry anchor point, wherein the transition segment is one of a straight line segment, a polyline segment, or a quadratic curve segment, and the direct splicing reference is obtained by splicing the reserved segment of the current reference path with the subsequent segment of the new path.

[0017] This invention also provides a global path update forward access reference continuation system for distributed weak tracking mobile platforms, the system being used to implement the method, comprising: The global planning layer is used to generate global paths and send path update packets to the path reference preprocessing layer. The path reference preprocessing layer, which is communicatively connected to the global planning terminal, includes: The state freeze unit is used to receive the path update packet and freeze the current pose, current velocity, current reference path, task generation, submitted anchor point index and local planning interface state within a processing cycle. The event classification and permission determination unit is used to classify the path update package according to the path validity, target change distance, path summary and task generation, and determine the status modification permission corresponding to each event type, forming a mapping from path update events to reference path status modification permissions. The access reference and advance radius calculation unit is used to use the current pose as the access reference and to calculate the optional advance radius based on the delivery delay statistics and the current speed when there is an observable delivery delay. The dynamic access boundary calculation unit is used to calculate the weak tracking deviation based on the minimum distance from the current pose to the unexecuted segment of the current reference path, and to calculate the dynamic access boundary based on the weak tracking deviation and the optional advance radius. The selection unit is used to choose between an exit anchor point or an entry anchor point based on the event classification results. The reference continuation output unit is used to output a direct splicing reference, a transition reference, or a degradation processing event based on the distance between the exit anchor point and the entry anchor point, the heading difference, and the obstacle risk. The interface and state management unit is used to output the reference path, event type, task generation, forward anchor index, spatial anchor position, exit anchor, entry anchor, dynamic access boundary, optional advance radius, degradation flag and path failure flag to the local planning end, and submit the current reference state or maintain the old reference based on the acceptance or rejection result of the local planning end. The local planning end is communicatively connected to the path reference preprocessing layer, and is used to receive the reference path and interface fields output by the path reference preprocessing layer, perform local trajectory optimization, and return an acceptance or rejection result to the interface and state management unit.

[0018] The present invention also provides an application of the global path update forward access reference continuation method for distributed weak tracking mobile platforms in robots.

[0019] Beneficial technical effects of the present invention By limiting the source of the problem through the context of distributed deployment, the technical contradictions between global path delay arrival, weak tracking deviation, and local reference mutation become clearer.

[0020] By unifying event classification and status modification permissions, keep-alive, rolling updates, new task switching, path invalidation, pending confirmation, and downgrade processing are mapped to different sets of modifiable statuses, thus avoiding the clearing of references as soon as the path package arrives.

[0021] By using both forward anchor index and spatial anchor position outputs, the local planning end can obtain an executable reference starting point while retaining the path progress, status commit, and log traceability basis.

[0022] By limiting this module to a reference path preprocessing layer before local planning, and by clearly defining its output geometric reference path and interface flags, it does not intrude on the local trajectory optimization, obstacle avoidance, and control command generation processes. Attached Figure Description

[0023] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings, wherein: Figure 1 A schematic diagram illustrating the location of the path reference preprocessing layer in a distributed deployment of existing technologies; Figure 2 This is a schematic diagram of the method flow of the present invention; Figure 3 This is a schematic diagram of the current pose, dynamic access boundary, and forward anchor point of the present invention; Figure 4 This is a timing diagram of the reference path output and local planning interface of the present invention. Detailed Implementation

[0024] To better understand the technical solution of this invention, the content of this invention includes, but is not limited to, the specific embodiments described below. Similar technologies and methods should be considered within the scope of protection of this invention. To make the technical problems to be solved, the technical solutions, and advantages of this invention clearer, a detailed description will be provided below in conjunction with the accompanying drawings and specific embodiments.

[0025] It should be understood that the embodiments described in this invention are merely some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without inventive effort are within the scope of protection of this invention.

[0026] The terminology used in the embodiments of this invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. The singular forms “a,” “the,” and “the” as used in the embodiments of this invention and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise.

[0027] This invention sets up a path reference preprocessing layer between the global planner and the local planner (close to the local planning end); the preprocessing layer classifies delayed global path updates into events, defines the state modification permissions of each event, uses the current pose of the local side to search for the foremost accessible anchor point within the dynamic access boundary, and simultaneously provides the local planning end with the reference path, anchor point index, spatial anchor point position and interface status flag.

[0028] like Figure 2 , Figure 3 and Figure 4 As shown, this invention discloses a global path update forward access reference continuation method for distributed weak tracking mobile platforms, comprising the following steps: S1. Receive the global path update packet and freeze the current pose, current velocity, current reference path, task generation, and local planning interface state; S2. Classify the events of the path update package according to the path validity, target change distance, path summary and task generation, and determine the status modification permissions corresponding to each event type, forming a mapping from path update events to reference path status modification permissions; S3. Use the current pose as the access reference, and calculate the optional advance radius based on the delivery delay and the current speed when there is an observable delivery delay; S4. Determine the weak tracking deviation based on the minimum distance from the current pose to the unexecuted segment of the current reference path, and calculate the dynamic access boundary based on the weak tracking deviation and the optional advance radius; S5. Select either exit anchor point or enter anchor point based on the event classification results; S6. Based on the distance between the exit anchor point and the entry anchor point, the heading difference, and the obstacle risk, output a direct splicing reference, a transition reference, or a downgraded processing event; S7. Output the reference path, event type, task generation, forward anchor index, spatial anchor position, exit anchor, entry anchor, dynamic access boundary, optional advance radius, degradation flag and path failure flag to the local planning end, and submit the current reference status or keep the old reference based on the acceptance or rejection result of the local planning end.

[0029] Preferably, the freezing in S1 also includes the submitted anchor point index and the most recent acceptance or rejection status of the local planning end; the freezing uses the same set of current pose, current velocity and current reference path within a processing cycle.

[0030] Preferably, the event classification in S2 includes keep-alive events, current task rolling update events, new task switching events, path failure events, pending confirmation events, and degradation processing events; the process of mapping path update events to reference path status modification permissions is as follows: keep-alive events allow modification of availability timestamps; rolling update events allow modification of unexecuted preceding reference paths and anchor fields; new task switching events allow modification of task generations and new reference entry points; path failure events allow modification of failure flags; pending confirmation events do not submit new references; degradation processing events allow modification of degradation flags and replanning requests.

[0031] Preferably, the optional advance radius in step S3 is determined as follows: the difference between the reference path sending time and the local planning end receiving time is recorded N times, the average value is taken to obtain the reference delivery delay, the reference delivery delay is subjected to amplitude limiting processing and multiplied by the modulus of the current speed to obtain the optional advance radius; if the delivery delay is unmeasurable, the jitter exceeds the preset threshold or the current speed is unavailable, the optional advance radius is zero, and N is a natural number.

[0032] Preferably, the dynamic access boundary in S4 is determined according to the following relationship: based on the basic access radius, the product of weak tracking deviation and deviation weight and the product of optional advance radius and advance radius weight are superimposed, and the calculation result is cropped to between the preset minimum access boundary and the maximum access boundary; wherein the weak tracking deviation is the minimum distance from the current pose to the unexecuted segment of the current reference path.

[0033] Preferably, S5 specifically includes: S51. When the event classification result is the current task is being updated, construct a candidate set on the new path that satisfies the forward constraint and whose distance does not exceed the sum of the dynamic access boundary and the optional advance radius. Select the point with the longest arc length progress from the candidate set as the forward anchor point, and output the forward anchor point index and the spatial anchor point position at the same time. S52. When the event classification result is a new task switch, select the exit anchor point in the unexecuted segment before the current reference path, and select the entry anchor point in the accessible segment before the new path.

[0034] Preferably, the exit anchor point in S52 is selected from the unexecuted segment preceding the current reference path; the entry anchor point is selected from the accessible segment preceding the new path.

[0035] Preferably, the method for constructing the transition reference in S7 is as follows: a transition segment is inserted between the exit anchor point and the entry anchor point, wherein the transition segment is one of a straight line segment, a polyline segment, or a quadratic curve segment, and the direct splicing reference is obtained by splicing the reserved segment of the current reference path with the subsequent segment of the new path.

[0036] Specifically, this invention uses a local planning preprocessing layer as an example. This local planning preprocessing layer is deployed before the local planning end and can run on the same machine, in the same process, or in the same control cycle as the local planning end. This module takes the global path update and the current robot state as input and outputs the real-time global reference path, anchor points, event types, degradation flags, and local planning interface constraints; the local planner is still responsible for local trajectory optimization, collision avoidance, and solving control constraints.

[0037] The technical boundary emphasized in this embodiment is that the local planning preprocessing layer only handles the reference continuation problem between the global path update and the local planning input, independent of global path or local trajectory production. This boundary facilitates the integration of this solution into existing distributed navigation systems with minimal modifications. Specific steps are as follows (e.g.) Figure 2 (as shown) S1: Path update packet reception and current state freeze The purpose of this step is to fix the same set of inputs within a processing cycle, so as to avoid inconsistencies in event classification, anchor point selection and state submission caused by changes in pose, velocity or old reference during the calculation process.

[0038] After the global planning module replans based on the actual situation and generates a path update, it sends the updated path as a path update packet to the path reference preprocessing layer via the communication link. The input path update packet includes the new global path P. new Target point g new Path summary ID path Path number seq path and path validity flag valid path When the path reference preprocessing layer receives the updated information, it immediately freezes the robot's current state, including the current pose x. now Current speed v r Current reference path or old reference path P old Current task objective g cur The current task generation and local planning interface status. The above update package inputs and the current status together constitute the computational input for this cycle.

[0039] Freezing the current state in this step means using the same set of current poses x within a processing cycle. now (Used as a reference for access and in distance calculation), current speed v r (Used to calculate the optional lead radius), current reference path P old (Used to calculate weak tracking deviation, unexecuted segments, and exit anchor points), current mission objective g cur(Used to calculate target change distance), task generation (used to distinguish between rolling updates and new task switching), interface status (used to record the most recent acceptance / rejection and decide whether to submit or keep the old reference), to avoid inconsistent judgments caused by status changes during anchor point selection.

[0040] S2: Event Classification and Status Modification Permission Determination The purpose of this step is to classify the events of the path update package based on path validity, target change distance, path summary, and task generation, and to determine the status modification permissions corresponding to each event type, thus forming a mapping from path update events to reference path status modification permissions.

[0041] After reading the path update package, analyze and process the variables within it: To distinguish between rolling updates of the same task and substantive task switching, the target change distance d is calculated. goal : d goal = ||g new -g cur || If the path is valid, the flag is valid. path An invalid path indicates a possible failure in global planning or a timeout in computation. Specifically, this might manifest as an empty path or a path being revoked, thus constituting a path failure event. If the new path digest ID... path If it remains unchanged and is only used to maintain link availability, it is a keep-alive event; if d goal ≤ε goal And if the task generation remains unchanged, then it is the current task's rolling update event, ε goal This is a manually set event update threshold, representing the tolerance for judging changes in the target; if d goal >ε goal If there is a generational change in the task, it is a new task switching event; if the path field is missing or the summary conflicts, it is a pending confirmation event; if the path is valid but the anchor point or switching cost does not meet the seamless access conditions, it is a degradation processing event. The classification results are used for access control, anchor point selection, direct splicing / transition / degradation output, and status submission.

[0042] Keep-alive events only update timestamps and do not modify reference paths or task generations; rolling update events allow replacing unexecuted preceding reference paths but prohibit progress rollback; new task switch events allow creating new task generations but prohibit reusing old task progress; path failure events output failure flags; pending confirmation events maintain the old reference; degradation processing events output replanning requests or low-speed access suggestions. In other words, different events have different state modification permissions, and changes in path summaries, targets, or validity will not automatically trigger the same reset action.

[0043] Preferably, the state modification permission is a mapping from event type to state set, specifically: keep-alive events only allow modification of availability time; rolling update events allow modification of unexecuted preceding reference and anchor fields; new task switch events allow modification of task generation and new reference entry; path failure events only allow modification of failure flags; pending confirmation events do not submit new references; and degradation processing events allow modification of degradation flags and replanning requests. This mapping enables path update processing to have auditable state boundaries.

[0044] S3: Access Baseline and Optional Lead Radius This step uses the current pose as the access reference and calculates an optional advance radius based on the delivery delay and the current velocity when an observable delivery delay exists. This step directly uses the current pose x. now This serves as the access benchmark for the preprocessing module. If this layer (preprocessing module) runs in the same cycle as the local planning end, the advance radius is set to zero.

[0045] If there is a delivery delay between the output of this layer and its acceptance by the local planning end, record the most recent N sending and receiving times. Take the average of the differences between the N reference path sending times and the local planning end receiving times to obtain the reference delivery delay. τ ref = avg N (t accept - t send ) In the formula, t send For the preprocessing layer to send a reference timestamp to the local planning end, t accept The timestamp of the acknowledgement returned by the local planning end is the difference between the timestamp and the timestamp, representing one observable delivery delay. N This represents the arithmetic mean of the most recent N valid send-receive samples, where N is a positive integer representing the number of samples.

[0046] To avoid excessive advance caused by delay jitter, the reference delivery delay is limited and then multiplied by the modulus of the current speed to obtain the advance radius R. adv : R adv = ||v r ‖clamp(τ ref , 0, τ max ) In the formula, τ max To allow for the inclusion of a pre-calculated maximum delivery delay limit, used to suppress abnormally long delays; clamp(τ) ref , 0, τ max ) indicates that τ ref Restriction (clipping) between 0 and τ max between.

[0047] This advance radius only affects the forward selection range of candidate anchor points and does not change the principle of using the current pose as the access reference. If the delay is unmeasurable, the jitter is too large, or the velocity input is unavailable, then R... adv Take zero.

[0048] This step converts the delivery delays caused by distributed communication and scheduling into an interpretable forward distance R. adv This ensures that the candidate anchor point is appropriately positioned in front of the robot, reducing the risk of the reference point falling behind after the delayed path is reached. When the delay is unmeasurable, jitter exceeds a threshold, or speed is unavailable, R is used. adv =0.

[0049] S4: Weak Tracking Deviation and Dynamic Access Boundary This step determines the weak tracking deviation based on the minimum distance from the current pose to the unexecuted segment of the current reference path, and calculates the dynamic access boundary based on the weak tracking deviation and the optional advance radius.

[0050] Weak tracking deviation is the minimum distance from the current pose to all points in the unexecuted segment of the current reference path, and its property is a scalar distance e. trk : e trk = min i ||x now - p i || In the formula, p i Let i be the i-th discrete path point in the unexecuted segment of the current reference path, i∈{0,1,...,M-1}, where M is the total number of path points contained in the unexecuted segment.

[0051] The dynamic access boundary is based on the basic access radius, superimposed with the product of weak tracking deviation and deviation weight, and the product of optional advance radius and advance radius weight, and the calculation result is cropped to between the preset minimum access boundary and maximum access boundary. D dyn = clamp(D0+ α·min(e trk ,e max ) + β·R adv D min D max ) Where D dyn D represents the dynamic access boundary actually used in this processing cycle. min and D max These are the preset minimum and maximum access boundaries, respectively. `clamp` indicates that the calculation results will be restricted (clipping) to D. min With D maxBetween. D0 is the base access radius, α is the deviation weight, and β is the advance radius weight. Both α and β are non-negative dimensionless weights, and their values ​​satisfy α≥0 and β≥0, with upper limits set by engineering calibration. e max This is the upper limit for abnormal deviations, set according to the specific engineering conditions.

[0052] This step enables the access range to adapt to the degree of weak tracking and communication delay, while preventing abnormal deviations from causing the access range to be too small or too large.

[0053] S5: Selection of forward anchor points for current task rolling updates The purpose of this step is: when the event classification result is the current task is being updated, construct a candidate set on the new path that satisfies the forward constraint and whose distance does not exceed the sum of the dynamic access boundary and the optional advance radius, select the point with the longest arc length progress from the candidate set as the forward anchor point, and output the forward anchor point index and the spatial anchor point position at the same time.

[0054] This step uses a new path, which is the path carried in this update package. After S2 confirms the path is valid, the task generation is consistent with the current task, and the target change distance does not exceed εgoal, it is used as the path to be accessed in the current task's rolling update. A candidate set is constructed on the new path, requiring that the candidate points in the set are located in front of the current pose, and the distance does not exceed the sum of the dynamic access boundary and the advance radius. Figure 3 As shown: A forward constraint is set, the judgment condition being: the dot product of the adjacent path direction vector of the candidate point and the candidate point to the current pose vector is not less than the negative value of the backward tolerance threshold; from the candidate set that satisfies the forward constraint and whose distance does not exceed the sum of the dynamic access boundary and the optional advance radius, the point with the largest arc length progress is selected as the forward anchor point, and the path index of the forward anchor point is used as the forward anchor point index, and the spatial coordinates of the forward anchor point are used as the spatial anchor point position, both output simultaneously: Q new = {q k | ‖q k -x now || ≤ D dyn + R adv , (q k+1 -q k )·(q k -x now ) ≥ -ε back} Where ε back It is the backward tolerance threshold, Q new To satisfy ||q k -x now || ≤ D dyn + R adv and the set of candidate points for forward constraints; qk Let k be the k-th discrete point on the new path; if the new path has N points in total, then k∈{0,...,N-1}.

[0055] The point with the earliest arc length progress in the candidate set is selected as the anchor point. This selection rule differs from the nearest point access: the nearest point only minimizes the spatial distance, while the forward anchor point prioritizes the point with the earliest arc length progress among the candidate points that meet the accessibility boundary, so as to reduce the probability of selecting a point behind the robot or a point that has already been passed when accessing a delayed path. k d * = arg max k s k , q k ∈Q new s k For q k The cumulative arc length progress along the starting point of the new path; q k ∈Q new This indicates that the maximum value operation is performed only within the candidate set filtered by forward pass and distance. d * It is a forward anchor index used to determine the starting path point, status commits, and log records of the output reference path.

[0056] Define q anchor For the corresponding spatial anchor point position, satisfying q anchor =q(k d * ), used as a reference starting point field for local planning.

[0057] The purpose of this step is to maintain the forward progress of the path arc length and avoid connecting to points behind or points that have already been passed.

[0058] S6: Forward exit and entry anchors for switching new tasks The purpose of this step is to select an exit anchor point in the unexecuted segment before the current reference path and an entry anchor point in the accessible segment before the new path when the event classification result is a new task switch.

[0059] For switching to a new task, remember i old For the old path P old The exit anchor point is numbered and its abbreviation is p. old Correspondingly, the p after that new It uses the same convention notation, exiting the anchor point p. old You must select from the unexecuted segments preceding the current reference path; you cannot select from anchor points that have already been passed. Enter anchor point q. new You must select a point that is accessible ahead of you in the new path; you cannot select a point behind you in the new path. Exit anchor point p.old Select by the following formula: p old = arg min i ||x now -p i ‖, satisfies i ≥ i now , and ||x now -p i || ≥ L pre Where L pre This is the minimum forward margin, used to ensure that the anchor point has pre-defined characteristics. now For the old path P that is currently being tracked old The index of the next point, in the above formula, means finding the point within the set of feasible points that has the smallest distance to the current pose, and the distance between this point and the current pose must satisfy the forward margin. The selection method for entering the anchor point is as follows: First, set L. max,0 To find the minimum intervention threshold, with initial values ​​set as fixed parameters, we search for new paths that satisfy ||p||. new -p old || ≤ L max The set of all points Q new,n’ = {q k’ | ‖q k’ -p old || ≤ L max,n} If Q new,0’ If it is an empty set, then repeat the iteration L. max,n+1 = L max,n + L delta until a non-empty set Q is found. new,n’ L delta This is the minimum increment unit, used to set the magnitude of the outward expansion of the search radius. When a non-empty set Q is found... new,n’ Then, find the point with the largest sequence number, which is the new forward access anchor point q. new k' is Q in the set new,n’ The index in the set Q, qk' is the set number. new,n’ The elements in.

[0060] This step ensures that both exiting and entering a new task switch has forward margin, avoiding backtracking of the old path, reversal of the new path, or unstable switching too close to the robot's current position.

[0061] S7: Direct splicing, transition reference, and degraded output This step outputs a direct splicing reference, transition reference, or degradation processing event based on the distance between the exit anchor point and the entry anchor point, the heading difference, and the obstacle risk. When the distance between exiting and entering the anchor point is small, the heading difference is small, and the obstacle risk is low (Q) new,0’ (Not empty), output directly concatenates the reference path P out : P out = P keep + P tail Where P keep For the old path P old down until exiting anchor point p old The previous path sequence, P tail For the new path Q new Downward access anchor point q new The subsequent path sequence, P out Combine the two in sequence.

[0062] When there is a distance or course discontinuity between the two but they can still be geometrically connected, i.e., Q new,n’ Non-empty, n ≥ 1, construct transition reference P bridge The simplest way to construct it is in p old With q new Insert a straight line segment between them; for a smoother finish, insert the midpoint and press p. old Midpoint, q new If a polyline or quadratic curve reference is generated, the output reference will be: P' out = P keep + P bridge + P tail Among them, P bridge This indicates a transitional reference point in the middle.

[0063] If the path is valid but seamless access is not currently possible, a degradation processing event will be output. Degradation processing includes degradation replanning requests and low-speed access suggestions, but is not the same as path failure.

[0064] This step transforms the anchor point geometry into an executable reference output: maintaining the shortest path connection when conditions are good, adding a smooth transition when there are discontinuities but connections are possible, and outputting degraded replanning or low-speed access suggestions when a safe and seamless access is not possible.

[0065] S8: Interface output, status commit, and rollback Based on the previous calculations, this step outputs P to the local planning end. out S2 classifies the event types obtained from path update packets. type The update package carries the task generation, which is then written into the interface field `task` after the preprocessing layer verifies the format and monotonicity. gen k d* Spatial anchor point position q anchor p old q new D dyn R adv Degradation flags and path failure flags, such as Figure 4 As shown. If the reference path is not empty, the events and tasks are valid in generations, the anchor index is consistent with the spatial location, the path connection satisfies the constraints such as boundary / collision / curvature, and a local trajectory can be generated within the planning cycle, the local planner returns acceptance and submits the current reference state; if any key field is missing, there is generational rollback, the anchor points are inconsistent, or the geometric / safety constraints are not met, the local planner returns rejection, and the old reference is maintained and the planner waits for the next path update.

[0066] Because of k d * and q anchor Simultaneously, the local planning end can either plan directly according to the reference path, or trace the reference path's starting point in debugging, visualization, rollback, status commit, or logs. This output belongs to the reference path interface and is independent of the output of local trajectories or control commands.

[0067] This step provides a complete, verifiable, and traceable reference path interface, and submits a new state with an accepted result and maintains the old reference with a rejected result, thus realizing the confirmation, submission, and rollback protection of distributed updates.

[0068] After executing the above steps, the global planning end generates an update package with path summary, validity, and task generation; the path reference preprocessing layer freezes the state, classifies events, compensates for delays, selects forward anchors, and determines the continuation mode; the local planning end receives the complete fields and generates the local trajectory. This achieves forward access, continuous continuation, committable / rollbackable state, and traceable logs for global path updates. The entire process reduces the risks of path rollback, reference mutations, and distributed timing jitter.

[0069] As an embodiment of the present invention, the present invention also provides a global path update forward access reference continuation system for distributed weak tracking mobile platforms, used to implement the method, including the following modules: The global planning layer is used to generate global paths and send path update packets to the path reference preprocessing layer. The path reference preprocessing layer, which is communicatively connected to the global planning terminal, includes: The state freeze unit is used to receive the path update packet and freeze the current pose, current velocity, current reference path, task generation, submitted anchor point index and local planning interface state within a processing cycle. The event classification and permission determination unit is used to classify the path update package according to the path validity, target change distance, path summary and task generation, and determine the status modification permission corresponding to each event type, forming a mapping from path update events to reference path status modification permissions. The access reference and advance radius calculation unit is used to use the current pose as the access reference and to calculate the optional advance radius based on the delivery delay statistics and the current speed when there is an observable delivery delay. The dynamic access boundary calculation unit is used to calculate the weak tracking deviation based on the minimum distance from the current pose to the unexecuted segment of the current reference path, and to calculate the dynamic access boundary based on the weak tracking deviation and the optional advance radius. The selection unit is used to choose between an exit anchor point or an entry anchor point based on the event classification results. The reference continuation output unit is used to output a direct splicing reference path, a transitional reference path, or a degradation processing event based on the distance between the exit anchor point and the entry anchor point, the heading difference, and the obstacle risk. The interface and state management unit is used to output the reference path, event type, task generation, forward anchor index, spatial anchor position, exit anchor, entry anchor, dynamic access boundary, optional advance radius, degradation flag and path failure flag to the local planning end, and submit the current reference state or maintain the old reference based on the acceptance or rejection result of the local planning end. The local planning end is communicatively connected to the path reference preprocessing layer, and is used to receive the reference path and interface fields output by the path reference preprocessing layer, perform local trajectory optimization, and return an acceptance or rejection result to the interface and state management unit.

[0070] The foregoing description illustrates and describes several preferred embodiments of the present invention. However, as previously stated, it should be understood that the present invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and environments, and can be altered within the scope of the inventive concept described herein through the foregoing teachings or techniques or knowledge in related fields. Any modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the present invention should be within the protection scope of the appended claims.

Claims

1. A global path update forward access reference continuation method for distributed weak tracking mobile platforms, characterized in that, Includes the following steps: S1. Receive the global path update packet and freeze the current pose, current velocity, current reference path, task generation, and local planning interface state; S2. Classify the events of the path update package according to the path validity, target change distance, path summary and task generation, and determine the status modification permissions corresponding to each event type, forming a mapping from path update events to reference path status modification permissions; S3. Use the current pose as the access reference, and calculate the optional advance radius based on the delivery delay and the current speed when there is an observable delivery delay; S4. Determine the weak tracking deviation based on the minimum distance from the current pose to the unexecuted segment of the current reference path, and calculate the dynamic access boundary based on the weak tracking deviation and the optional advance radius; S5. Select either exit anchor point or enter anchor point based on the event classification results; S6. Based on the distance between the exit anchor point and the entry anchor point, the heading difference, and the obstacle risk, output a direct splicing reference path, a transitional reference path, or a downgraded processing event; S7. Output the reference path, event type, task generation, forward anchor index, spatial anchor position, exit anchor, entry anchor, dynamic access boundary, optional advance radius, degradation flag and path failure flag to the local planning end, and submit the current reference status or keep the old reference based on the acceptance or rejection result of the local planning end.

2. The method according to claim 1, characterized in that, The freezing described in S1 also includes the submitted anchor point index and the most recent acceptance or rejection status of the local planning end; the freezing uses the same set of current pose, current velocity and current reference path within a processing cycle.

3. The method according to claim 1, characterized in that, The event categories described in S2 include keep-alive events, current task rolling update events, new task switching events, path failure events, pending confirmation events, and degradation processing events; The process of mapping path update events to reference path state modification permissions is as follows: keep-alive events allow modification of availability timestamps; The rolling update event allows modification of the unexecuted preceding reference path and anchor point fields; the new task switch event allows modification of the task generation and the new reference entry. Path failure events allow modification of failure flags; Pending events do not submit new references; downgraded events allow modification of downgrade flags and replanning requests.

4. The method according to claim 1, characterized in that, The method for determining the optional advance radius in step 3 is as follows: record the difference between the reference path sending time and the local planning end receiving time N times, take the average value to obtain the reference delivery delay, and multiply the reference delivery delay by the modulus of the current speed after the amplitude limiting processing to obtain the optional advance radius; If the delivery delay is unmeasurable, the jitter exceeds a preset threshold, or the current speed is unavailable, the optional advance radius is set to zero, and N is a natural number.

5. The method according to claim 1, characterized in that, The dynamic access boundary described in S4 is determined according to the following relationship: based on the basic access radius, the product of weak tracking deviation and deviation weight and the product of optional advance radius and advance radius weight are superimposed, and the calculation result is cropped to between the preset minimum access boundary and the maximum access boundary; wherein the weak tracking deviation is the minimum distance from the current pose to the unexecuted segment of the current reference path.

6. The method according to claim 1, characterized in that, S5 specifically includes: S51. When the event classification result is the current task is rolling update, construct a candidate set on the new path that satisfies the forward constraint and whose distance does not exceed the sum of the dynamic access boundary and the optional advance radius. Select the point with the longest arc length progress from the candidate set as the forward anchor point, and output the forward anchor point index and spatial anchor point position at the same time. S52. When the event classification result is a new task switch, select the exit anchor point in the unexecuted segment before the current reference path, and select the entry anchor point in the accessible segment before the new path.

7. The method according to claim 6, characterized in that, The exit anchor point described in S52 is selected from the unexecuted segments preceding the current reference path; the entry anchor point is selected from the accessible segments preceding the new path.

8. The method according to claim 1, characterized in that, The method for constructing the transition reference path described in S6 is as follows: a transition segment is inserted between the exit anchor point and the entry anchor point. The transition segment is one of a straight line segment, a polyline segment, or a quadratic curve segment. The direct splicing reference path is obtained by splicing the reserved segment of the current reference path with the subsequent segment of the new path.

9. A global path update forward access reference continuation system for distributed weak tracking mobile platforms, characterized in that, The system is used to implement the method according to any one of claims 1-8, comprising: The global planning layer is used to generate global paths and send path update packets to the path reference preprocessing layer. The path reference preprocessing layer, which is communicatively connected to the global planning terminal, includes: The state freeze unit is used to receive the path update packet and freeze the current pose, current velocity, current reference path, task generation, submitted anchor point index and local planning interface state within a processing cycle. The event classification and permission determination unit is used to classify the path update package according to the path validity, target change distance, path summary and task generation, and determine the status modification permission corresponding to each event type, forming a mapping from path update events to reference path status modification permissions. The access reference and advance radius calculation unit is used to use the current pose as the access reference and to calculate the optional advance radius based on the delivery delay statistics and the current speed when there is an observable delivery delay. The dynamic access boundary calculation unit is used to calculate the weak tracking deviation based on the minimum distance from the current pose to the unexecuted segment of the current reference path, and to calculate the dynamic access boundary based on the weak tracking deviation and the optional advance radius. The selection unit is used to choose between an exit anchor point or an entry anchor point based on the event classification results. The reference continuation output unit is used to output a direct splicing reference path, a transitional reference path, or a degradation processing event based on the distance between the exit anchor point and the entry anchor point, the heading difference, and the obstacle risk. The interface and state management unit is used to output the reference path, event type, task generation, forward anchor index, spatial anchor position, exit anchor, entry anchor, dynamic access boundary, optional advance radius, degradation flag and path failure flag to the local planning end, and submit the current reference state or maintain the old reference based on the acceptance or rejection result of the local planning end. The local planning end is communicatively connected to the path reference preprocessing layer, and is used to receive the reference path and interface fields output by the path reference preprocessing layer, perform local trajectory optimization, and return an acceptance or rejection result to the interface and state management unit.

10. The application of the global path update forward access reference continuation method for distributed weak tracking mobile platforms as described in any one of claims 1-8 in a robot.

Images