本数据集面向智能交通系统中单智能体最短路径规划算法的性能评估需求，其生成背景源于城市道路网络中机器人或车辆在静态障碍物环境下的路径搜索与导航问题。数据集基于巴黎某城区的网格化扫描地图构建（详见R. Stern et al., \"Multi-agent pathfinding: Definitions, variants, and bench-marks,\" in Proc. Int. Symp. Combinatorial Search, vol. 10, no. 1, 2019,pp. 151-158.），采用256×256栅格表示，其中黑色区域为建筑障碍物（不可通行），白色区域为可通行区域。将每个栅格点视为网络节点，可通行区域内的节点与上、下、左、右、左上、左下、右上、右下八个方向的相邻节点相连，形成移动路径，障碍物节点则隔离于网络之外。由此构建的大规模路径网络包含65536个节点和354938条边，得到65536*8的稀疏邻接矩阵数据文件Paris_1_256_weight_nodirect.mat以及其局部有向图矩阵数据Paris_1_256_weight_direct.mat，为算法测试提供了高复杂度场景。本数据集支持在无向图和有向图（通过限制部分节点的邻居方向）两种模式下进行路径规划实验，已用于验证多种离散时间布谷鸟搜索（DBMC）控制策略的收敛性能。数据内容包含一个完整的网格地图网络拓扑（gml或其他格式），以及明确的起终点节点标识（起点为节点65374，终点为节点1），可有效支撑路径规划算法的最短路径求解能力、收敛速度及鲁棒性研究。
", "ds_source": "本数据集基于仿真生成方式构建，其原始地图来源于巴黎某城区的实际扫描地图，通过栅格化处理得到256×256分辨率的二值网格（黑色为障碍，白色为可通行区域）。地图数据未直接来自特定文献或公开数据集下载，而是基于该扫描图进行网格化加工生成。所有节点和边的拓扑结构完全依据网格黑白属性及邻接规则自动生成，不包含实测交通流量或动态环境数据。
", "ds_process_way": "本数据集通过编程方式加工生成。首先将巴黎某城区的扫描地图转换为256×256的二值网格图像，其中黑色像素映射为障碍节点（不可通行），白色像素映射为可通行节点。然后，以每个网格点作为网络节点，对于可通行区域内的节点，建立与其八个方向（上、下、左、右、左上、左下、右上、右下）相邻可通行节点的无向边；对于障碍节点，不建立任何边。最后设置边权重：垂直和水平方向的边权重为1，对角线方向的边权重为√2（保留浮点数精度）。接着，对于有向图模式，选取矩形区域，将该区域内节点的邻居数从八个缩减为三个（仅保留右、右上、右下方向），从而将无向图转化为有向图。最终导出65536*8的稀疏邻接矩阵数据文件Paris_1_256_weight_nodirect.mat以及其局部有向图矩阵数据Paris_1_256_weight_direct.mat。
", "ds_quality": "本数据集在生成过程中实施了严格的质量控制，确保了数据的完整性、准确性和可用性。经核查，网格地图尺寸严格为256×256，节点总数65536，未出现节点缺失或编号跳变；边总数354938，每条边均正确关联两个可通行节点，且权重取值准确（1或√2），无异常值或重复边。障碍区域与可通行区域的二值划分清晰，不存在模糊像素。图连通性检查表明，整个可通行区域构成一个弱连通无向图（在有向图模式下，修改区域内的连通性符合预期），确保任意两点间存在可行路径。起终点节点（65 374和1）均位于可通行区域，且存在至少一条最短路径。通过与经典Dijkstra算法基准对比，验证了所有边权和邻接关系的正确性。该数据集已成功用于多种路径规划算法的对比测试，实验结果（如最短路径长度、收敛曲线）可重复且稳定，表明数据质量可靠，可有效支撑路径规划算法的评估与比较。
", "ds_acq_start_time": "2025-01-01 00:00:00", "ds_acq_end_time": null, "ds_acq_place": "巴黎", "ds_acq_lon_east": null, "ds_acq_lat_south": null, "ds_acq_lon_west": null, "ds_acq_lat_north": null, "ds_acq_alt_low": null, "ds_acq_alt_high": null, "ds_share_type": "open-access", "ds_total_size": 190996, "ds_files_count": 0, "ds_format": "mat", "ds_space_res": "", "ds_time_res": "", "ds_coordinate": "无", "ds_projection": "", "ds_thumbnail": "3fcf28f7-547c-4cfe-826e-3c6589802fff.png", "ds_thumb_from": 0, "ds_ref_way": "", "paper_ref_way": "", "ds_ref_instruction": "", "ds_from_station": "", "organization_id": "9ecaaa78-39e9-411e-9f24-274e12aa643f", "ds_serv_man": "莫远秋", "ds_serv_phone": "13216110989", "ds_serv_mail": "yuanqiumo@seu.edu.cn", "doi_value": "", "subject_codes": [ "410" ], "quality_level": 0, "publish_time": "2026-07-09 10:58:19", "last_updated": "2026-07-09 10:58:19", "protected": false, "protected_to": "2028-06-30 00:00:00", "lang": "zh", "cstr": "", "i18n": { "en": { "title": "Grid-based Path Planning Simulation Dataset for Intelligent Transportation Networks", "ds_format": "mat", "ds_source": "This dataset is constructed based on simulation-based generation. Its original map is derived from an actual scanned map of a Parisian district, which is processed through rasterization to obtain a 256×256 binary grid (with black regions representing obstacles and white regions representing traversable areas). The map data are not directly sourced from specific literature or public dataset downloads, but are generated by gridding and processing the scanned map. The topological structure of all nodes and edges is fully automatically generated based on the binary attributes of the grid and the adjacency rules, and does not contain measured traffic flow or dynamic environment data.", "ds_quality": "During the generation of this dataset, rigorous quality control measures were implemented to ensure its completeness, accuracy, and usability. Upon verification, the grid map dimensions are strictly 256×256, with a total of 65,536 nodes, and no missing nodes or ID discontinuities were observed. The total number of edges is 354,938; each edge correctly connects two traversable nodes, and the edge weights are accurately assigned (either 1 or √2), with no outliers or duplicate edges. The binary classification between obstacle regions and traversable areas is clear, with no ambiguous pixels. Connectivity checks confirm that the entire traversable region forms a weakly connected undirected graph (and in the directed graph mode, the connectivity within the modified region meets expectations), ensuring that a feasible path exists between any two nodes. Both the start and destination nodes (65,374 and 1) are located in traversable areas, and at least one shortest path exists between them. The correctness of all edge weights and adjacency relationships was validated by comparison with the benchmark Dijkstra algorithm. This dataset has been successfully used in comparative tests of various path planning algorithms, and the experimental results (e.g., shortest path lengths, convergence curves) are reproducible and stable, demonstrating that the data quality is reliable and can effectively support the evaluation and comparison of path planning algorithms.", "ds_ref_way": "", "ds_abstract": "This dataset is designed for the performance evaluation of single-agent shortest path planning algorithms in intelligent transportation systems. Its generation background originates from path search and navigation problems of robots or vehicles in urban road networks under static obstacle environments. The dataset is constructed based on a grid‑based scanned map of a Parisian district (see R. Stern et al., \"Multi-agent pathfinding: Definitions, variants, and benchmarks,\" in Proc. Int. Symp. Combinatorial Search, vol. 10, no. 1, 2019, pp. 151–158), using a 256 × 256 grid representation, where black regions denote building obstacles (impassable) and white regions denote traversable areas. Each grid point is treated as a network node. Nodes in traversable areas are connected to their eight adjacent neighbors in the directions of up, down, left, right, upper‑left, lower‑left, upper‑right, and lower‑right, forming movement paths, while obstacle nodes are isolated from the network. The resulting large‑scale path network contains 65,536 nodes and 354,938 edges, yielding a 65,536 × 8 sparse adjacency matrix data file named Paris_1_256_weight_nodirect.mat and its corresponding local directed graph matrix file Paris_1_256_weight_direct.mat, providing a high‑complexity scenario for algorithm testing. This dataset supports path planning experiments in both undirected and directed graph modes (by restricting neighbor directions of some nodes), and has been used to verify the convergence performance of various discrete‑time cuckoo search (DBMC) control strategies. The data content includes a complete grid map network topology (in GML or other formats) and clear start and end node identifiers (start node 65,374 and end node 1), which can effectively support research on shortest path solving capability, convergence speed, and robustness of path planning algorithms.", "ds_time_res": "", "ds_acq_place": "Nanjing", "ds_space_res": "", "ds_projection": "", "ds_process_way": "This dataset was generated through programming. First, the scanned map of a Parisian district was converted into a 256×256 binary grid image, where black pixels were mapped as obstacle nodes (untraversable) and white pixels as traversable nodes. Then, each grid point was treated as a network node; for nodes in traversable areas, undirected edges were established to their adjacent traversable nodes in eight directions (up, down, left, right, upper‑left, lower‑left, upper‑right, and lower‑right), while obstacle nodes were assigned no edges. Subsequently, edge weights were set as follows: vertical and horizontal edges had a weight of 1, and diagonal edges had a weight of √2 (with floating‑point precision preserved). Next, for the directed graph mode, a rectangular region was selected, and the number of neighbors for nodes within this region was reduced from eight to three (retaining only the right, upper‑right, and lower‑right directions), thus converting the undirected graph into a directed one. Finally, the sparse adjacency matrix data files were exported: a 65536×8 matrix file named Paris_1_256_weight_nodirect.mat for the full undirected graph, and a local directed graph matrix file named Paris_1_256_weight_direct.mat.", "ds_ref_instruction": "" } }, "submit_center_id": "ncdc", "data_level": 0, "recommendation_value": 0, "license_type": "https://creativecommons.org/licenses/by/4.0/", "doi_reg_from": "reg_local", "cstr_reg_from": "reg_local", "doi_not_reg_reason": null, "cstr_not_reg_reason": null, "is_paper_in_submitting": false, "belong_to_nieer": false, "ds_topic_tags": [ "智能交通", "路径规划" ], "ds_subject_tags": [ "工程与技术科学基础学科" ], "ds_class_tags": [], "ds_locus_tags": [], "ds_time_tags": [], "ds_contributors": [ { "true_name": "莫远秋", "email": "yuanqiumo@seu.edu.cn", "work_for": "东南大学", "country": "中国" } ], "ds_meta_authors": [ { "true_name": "莫远秋", "email": "yuanqiumo@seu.edu.cn", "work_for": "东南大学", "country": "中国" } ], "ds_managers": [ { "true_name": "莫远秋", "email": "yuanqiumo@seu.edu.cn", "work_for": "东南大学", "country": "中国" } ], "category": "其他" }