Path planning method based on regret minimization
By constructing a multi-dimensional attribute path graph and a regret minimization strategy, the Skyline path set is dynamically updated, solving the problem of combining real-time traffic data and user preferences in path planning. This achieves efficient personalized path recommendation and reduces computational complexity and regret value.
Patent Information
- Application Number
- CN202511820367.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-04
- Publication Date
- 2026-02-10
AI Technical Summary
Existing route planning techniques struggle to adapt to the dynamic changes in real-time traffic data in multi-objective query scenarios. Path pruning strategies are not adequately integrated with user preferences, leading to discrepancies between recommended results and actual needs. Furthermore, the high computational complexity makes it difficult to achieve efficient and personalized route recommendations.
By constructing a multi-dimensional attribute path graph, a Skyline path set is generated. Based on the regret minimization strategy, the utility function of the path is calculated, the maximum regret rate is minimized, and the recommendation set is dynamically updated to ensure that the path has a low regret value under various preference scenarios.
It enables efficient personalized route recommendation based on real-time traffic data, improves computational efficiency, ensures that recommended routes match user preferences, and reduces the complexity and regret value of user choices.
Smart Images
Figure CN121498735A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of path planning technology, and in particular to a path planning method based on regret minimization. Background Technology
[0002] With the rapid development of mobile internet and location services, route planning technology has become a core requirement in fields such as intelligent navigation and travel recommendation. In multi-objective route query scenarios, users often face trade-offs between multiple optimization objectives (such as time, distance, and cost), and their preferences may be ambiguous or dynamically changing. Traditional Skyline route query methods provide basic decision support for users by filtering out a set of candidate routes that are not mutually dominant across multiple objective dimensions. However, due to the complex trade-offs between route attributes, the Skyline result set is usually enormous, making it difficult for users to efficiently select representative routes that match their potential preferences.
[0003] Existing technologies primarily compress the size of candidate paths through vector weighting models or convex preference space modeling, such as greedy path selection or preference region partitioning methods. While these methods reduce computational complexity to some extent, they still have significant limitations: First, most methods rely on static graph assumptions, making it difficult to adapt to the dynamic changes in real-time traffic data (such as sudden congestion); second, path pruning strategies lack systematic integration with representativeness metrics, potentially overlooking high-quality paths that users might be interested in; third, existing optimization frameworks do not fully consider the dynamic modeling of user preference uncertainty, leading to discrepancies between recommended results and actual needs. These problems limit the effectiveness of path planning technology in real-time, personalized scenarios.
[0004] Therefore, a path planning method based on regret minimization is needed. Summary of the Invention
[0005] In view of this, the present invention provides a path planning method based on regret minimization, which measures the worst performance loss of the selected path in the candidate path set under unknown preferences by using the regret minimization strategy; by minimizing the maximum regret rate, it ensures that the returned path set has a low "regret value" under various preference scenarios, thereby achieving path recommendation.
[0006] Therefore, the present invention provides the following technical solution: A path planning method based on regret minimization includes: Based on real road network information and a set of various attribute points on the road network, a path map with multi-dimensional attributes is constructed. Based on the path graph, generate a set of paths from the starting point to all points of interest, and determine the Skyline path set based on the path set according to the comprehensive cost function. Calculate the utility function for each path in the Skyline path set, and determine the recommendation set by minimizing the maximum regret rate based on the utility function.
[0007] Further, the calculation of the utility function for each path in the Skyline path set, and the selection of the recommendation set based on minimizing the maximum regret rate according to the utility function, includes: During the route planning process, the Skyline route set is updated in real time based on the starting point's location. For a new route in the Skyline route set, the utility value of the new route is calculated. When the difference between the utility value of the new route and the route with the largest utility value in the recommended routes is greater than 0, and the utility value of the new route is greater than a preset threshold, the recommended route set is updated.
[0008] Furthermore, the updated recommendation set includes: If the recommendation set is not full, the new path will be added to the recommendation set; If the recommendation set is full, the path with the lowest path score in the recommendation set is deleted, and the new path is added to the recommendation set.
[0009] Furthermore, the path diagram includes:
[0010] in, This represents a set of nodes, including all intersections. Represents a set along the roadside; Represent each edge Assigned a dimension as The attribute vector includes the following edge attributes: road length, traffic time, toll, average congestion index, number of traffic lights, and number of intersections.
[0011] Further, determining the Skyline path set based on the path set according to the comprehensive cost function includes: Each path has a corresponding comprehensive cost function; For each path, determine whether there are other paths that are better than or equal to the path in the comprehensive cost function across all dimensions; If no other path exists, the candidate path is added to the Skyline path set.
[0012] Furthermore, the comprehensive cost index includes:
[0013] in, Represents the overall cost function; Representing a path The cumulative attribute vector; Represents the vector of attributes of points of interest. The attribute values represent points of interest, and m represents the total number of attribute values.
[0014] Further, the preset threshold is determined based on the regret value, including:
[0015] in, Regret value These are preset parameters. path In preferences The utility value is R; R is the recommendation set.
[0016] Furthermore, the utility function includes:
[0017] in, path In preferences Lower utility value, This represents the inner product of the preference vector and the path attribute. Representing a path In the The values of each attribute.
[0018] Advantages and positive effects of the present invention: This method combines interest recommendation and path optimization, simultaneously providing users with their points of interest and the optimal path to those points. It improves computational efficiency by selecting the Skyline path set using a cost set, and selects the recommendation set using regret values, thus satisfying both user preferences and the selection of the optimal path. Attached Figure Description
[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0020] Figure 1 This is a flowchart of the path planning method based on regret minimization in an embodiment of the present invention; Figure 2 This is a logic diagram of the path planning method based on regret minimization in an embodiment of the present invention; Figure 3 This is a schematic diagram of the road network in an embodiment of the present invention; Figure 4 This is a diagram showing the actual distance results of the path planning method based on regret minimization in an embodiment of the present invention. Detailed Implementation
[0021] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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 should fall within the scope of protection of the present invention.
[0022] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0023] This invention provides a path planning method based on regret minimization. It constructs a path graph with multi-dimensional attributes based on real road network information and a set of various attribute points on the road network. A Skyline path set from the current starting point to the user-provided points of interest is constructed based on the current starting point and user-provided points of interest. Based on minimizing the maximum regret rate, points of interest are recommended, and the Skyline path set is combined to output the points of interest and the optimal recommended path.
[0024] Example Combination Figure 1 As shown, the path planning method based on regret minimization includes: S1. Based on the real road network information and the set of various attribute points on the road network, construct a path map with multi-dimensional attributes.
[0025] Real road network information includes: the set of road edges and the attribute dimensions of each edge.
[0026] The attribute dimensions of each edge include: road length, traffic time, toll, average congestion index, number of traffic lights, and number of intersections; The point of interest set includes all hotels within the preset range.
[0027] Define a multidimensional attribute path graph
[0028] in, The node set represents all intersections; Represents a set along the roadside; Represent each edge Assigned a dimension as The attribute vector includes the following edge attributes: road length, traffic time, toll, average congestion index, number of traffic lights, and number of intersections.
[0029] S2. Based on the current starting point and the user-provided points of interest, construct a set of Skyline paths from the starting point to the points of interest.
[0030] 1. Based on a multi-dimensional attribute path graph, simulate all possible paths a user might take, and construct a path cost set, expressed by the formula:
[0031] in, Representing a path The cumulative attribute vector; Represents a node To the node The corresponding edge attribute vector; Indicates the first One intersection; Indicates the first Dimensional attributes in path The sum of the above; This indicates the total number of intersections on the path. This indicates the intersection index.
[0032] In this embodiment, the path attributes include: road length, travel time, toll, average congestion index, number of traffic lights, and number of intersections.
[0033] 2. Construct a comprehensive cost function considering user preferences and path.
[0034] The comprehensive cost function is obtained by combining user preference cost and path cost, and the formula is expressed as:
[0035] in, Represents the overall cost function; Representing a path The cumulative attribute vector; Represents the vector of attributes of points of interest. The attribute values represent points of interest, and m represents the total number of attribute values.
[0036] Determine the dominance conditions between paths: If path Domination Path ,Right now Then it must satisfy: 1) It is no worse than in all attribute dimensions The formula is expressed as:
[0037] 2) It is better in at least one attribute, as expressed by the formula:
[0038] Based on preset dominance judgment path conditions, determine the set of paths to Skyline:
[0039] in, Indicates starting from the origin To point of interest The Skyline path collection, Let represent the set of paths from the starting point s to all points of interest that meet the criteria.
[0040] Determine the starting point To the set of all points of interest:
[0041] in, Indicates starting from the origin The set of paths to all points of interest; N is the total number of Skyline paths from the starting point to all points of interest, and r represents the number of points of interest.
[0042] S3, Combination Figure 2 As shown, regret values are introduced to prune the Skyline path set and obtain a recommended set.
[0043] 1. Construct a comprehensive cost function considering user preferences and path.
[0044] The comprehensive cost function is obtained by combining user preference cost and path cost, and the formula is expressed as:
[0045] in, Represents the overall cost function; Representing a path The cumulative attribute vector; Represents the vector of attributes of points of interest. The attribute values represent points of interest, and m represents the total number of attribute values.
[0046] 2. Calculate the regret value based on user preferences, and determine the recommendation set based on the regret value.
[0047] 1) Construct a preference space and sample preference vectors, which are the utility functions of users' psychological needs. To simulate the degree of attention paid to path attributes by diverse users, define a set of user preference vectors:
[0048] in, This represents the preference vector, i.e., the user's edge attribute space. The attributes include: road length, travel time, toll fees, average congestion index, number of traffic lights, and number of intersections, with varying weights. This represents the sum of path attributes and point-of-interest attributes.
[0049] To ensure the meaningful comparison of preference vectors, each preference vector must be a unit vector, i.e., satisfying:
[0050] The preference space is constructed by applying the first quadrant (i.e., all components are non-negative) to... Unit sphere in space By performing uniform sampling, a set of normalized preference vectors is obtained:
[0051] Where k represents the number of dimensions, which is also the number of attributes.
[0052] For any path In preferences The definition of lower utility is:
[0053] in, path In preferences Lower utility value, This represents the inner product of the preference vector and the path attribute. Representing a path In the The values of each attribute.
[0054] 2) Figure 3 As shown, the user's current starting point is q, and the points of interest D1, D2, and D3 are determined based on the user's input target, thus constructing a road network. Starting from the first intersection A, an initial judgment is made based on the aforementioned dominance judgment to obtain the Skyline path set; the utility value of each path in the Skyline path set is calculated, and the regret value is calculated based on the utility value; the recommendation set is determined based on the regret value.
[0055] The definition of regret value is:
[0056] in, In the recommended collection The path with the highest utility is the best path provided to the user. This function represents the meaning of the newly generated path. According to the recommended set The path with the highest utility and new paths The difference is used to determine whether to update the recommendation set: If recommended collection If the difference between the path with the highest utility and the new path is 0, then the recommendation set already contains the equivalent path, and there is no need to update the recommendation set. If recommended collection If the difference between the path with the highest utility and the new path is greater than a preset threshold, then there is a defect in the recommendation set. The new path is then added to the recommendation set, and the recommendation set is updated. Define a dynamic threshold based on the regret value of the new path, and determine whether to add [the path] based on the threshold. The threshold formula for including a product in the recommendation set is:
[0057] A path is only included in the recommendation set when it significantly improves the user's potential utility; that is, when... Only then will the new path be added to the recommendation set R.
[0058] 3) Calculate the score of each path in the recommendation set, sort the paths in the recommendation set according to the score, and replace the path with the lowest score with a new path.
[0059] The size of the recommendation set R is limited because it is provided for users to choose from, so it cannot be infinitely large. Therefore, a scoring function is defined for each recommended path in set R to determine which path to replace in the current set. The formula is expressed as:
[0060]
[0061] in, Score path P. ; The contribution value of path P; the score function represents the path's contribution value. At what number of preference vectors is the utility value higher than the current recommendation set? The utility of all paths in the set is compared using a scoring function, and then the recommendations are inserted or deleted accordingly.
[0062] 4) When the path is subsequently expanded, the regret value may decrease.
[0063] Deleted paths are stored in a lazy pool, with a delayed deletion strategy. A lazy pool is dynamically maintained, adding paths with slightly lower regret values that are removed from the recommendation system and temporarily dominated. These paths are expanded along with the system during expansion. The lazy pool is not infinitely large; it's impossible to store all paths, otherwise pruning would be meaningless. Several management strategies are introduced for the lazy pool: Limit the time of existence Add the current path to the lazy pool and set its remaining lifetime. After each expansion The path will be deleted when the time to live is 0.
[0064] After expansion, the score is recalculated, and the path with the lowest score is deleted when the lazy pool exceeds the limit.
[0065] Only at this point is the calculation of the path to point A in Figure d complete. Then, we continue to expand point A, connect it with its neighbors B and C, perform another dominance judgment, and then connect it with the point of interest to repeat the above steps.
[0066] Once all path nodes have been expanded, the final recommendation set R is returned to the user for selection.
[0067] The actual distance result diagram based on the regret minimization path planning method in this embodiment is shown below. Figure 4 As shown.
[0068] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.
Claims
1. A path planning method based on regret minimization, characterized in that, include: Based on real road network information and a set of various attribute points on the road network, a path map with multi-dimensional attributes is constructed. Based on the path graph, generate a set of paths from the starting point to all points of interest, and determine the Skyline path set based on the path set according to the comprehensive cost function. Calculate the utility function for each path in the Skyline path set, and determine the recommendation set by minimizing the maximum regret rate based on the utility function.
2. The method according to claim 1, characterized in that, The calculation of the utility function for each path in the Skyline path set, and the selection of the recommendation set based on minimizing the maximum regret rate according to the utility function, includes: During the route planning process, the Skyline route set is updated in real time based on the starting point's location. For a new route in the Skyline route set, the utility value of the new route is calculated. When the difference between the utility value of the new route and the route with the largest utility value in the recommended routes is greater than 0, and the utility value of the new route is greater than a preset threshold, the recommended route set is updated.
3. The method according to claim 2, characterized in that, The updated recommendation set includes: If the recommendation set is not full, the new path will be added to the recommendation set; If the recommendation set is full, the path with the lowest path score in the recommendation set is deleted, and the new path is added to the recommendation set.
4. The method according to claim 1, characterized in that, The path diagram includes: in, This represents a set of nodes, including all intersections. Represents a set along the roadside; Represent each edge Assigned a dimension as The attribute vector includes the following edge attributes: road length, traffic time, toll, average congestion index, number of traffic lights, and number of intersections.
5. The method according to claim 1, characterized in that, The step of determining the Skyline path set based on the path set according to the comprehensive cost function includes: Each path has a corresponding comprehensive cost function; For each path, determine whether there are other paths that are better than or equal to the path in the comprehensive cost function across all dimensions; If no other path exists, the candidate path is added to the Skyline path set.
6. The method according to claim 1, characterized in that, The comprehensive cost index includes: in, Represents the overall cost function; Representing a path The cumulative attribute vector; Represents the vector of attributes of points of interest. The attribute values represent points of interest, and m represents the total number of attribute values.
7. The method according to claim 2, characterized in that, The preset threshold is determined based on the regret value, including: in, Regret value These are preset parameters. path In preferences The utility value is R; R is the recommendation set.
8. The method according to claim 1, characterized in that, The utility function includes: in, path In preferences Lower utility value, This represents the inner product of the preference vector and the path attribute. Representing a path In the The value of each attribute.