结合虚拟测试验证要点，梳理测试场景构成要素，实现场景要素跨域融合，构建测试船、目标船和环境间的博弈与对抗行为特征，针对场景表示的3种抽样层次，提出面向虚拟测试的场景生成方法。基于场景优化生成方法，在场景组合生成方法的基础上离线生成测试场景，主要针对参数空间的边缘场景，让智能算法参与场景生成过程，其场景覆盖度更低，通过测试高风险的边界场景，以测试特定系统性能效果。以单船避碰为例，在虚拟仿真测试平台进行仿真验证，结果表明优化生成后的场景文件能够有效降低场景生成数量，显著提升生成效率，且在场景交互博弈、覆盖率和可重复测试等方面具有良好性能。
Combined with the key points of intelligent ship testing and verification, analyze the components of high-risk edge test scenarios, realize cross-domain fusion of edge scenario elements in the virtual simulation test system, simulate the game and confrontation behavior characteristics between the ship and other ships and the environment in the real world, and express the scene for the scene. The three sampling levels are proposed, and a virtual test-oriented edge scene construction system is proposed. The scene keywords are extracted based on the semantic model, mapped to the parameter space to generate dynamically changing logical scenes, the scene model is described as a scene dynamics model, the scene controller is constructed by using a general function approximator, and realize the optimal solution of edge scene controller with the help of reinforcement learning. Finally, the parameter space is sampled by keywords to generate specific edge scene files. Taking the simultigle-vessel encounter scene of a test site as an example, the simulation verification is carried out on the virtual simulation test platform. The results show that the generated edge scene file can meet the requirements of the automatic driving test of the ship, significantly improve the generation efficiency. And it has good performance in scene interactive game, coverage and repeatable testing, etc.
2024,46(1): 164-169 收稿日期：2022-09-15
DOI：10.3404/j.issn.1672-7649.2024.01.028
分类号：U692.5+1
作者简介：苗雨阳(1998-),男,硕士研究生,研究方向为智能船舶测试评估
参考文献：
[1] 刘佳仑, 杨帆, 马吉林, 等. 智能船舶及测试验证评估技术研究现状与展望[J]. 中国船检, 2021(7): 68–74
LIU Jia-lun, YANG Fan, MA Ji-lin, et al. Research status and prospect of intelligent ship and test verification and evaluation technology[J]. China Ship Inspection, 2021(7): 68–74
[2] 刘佳仑, 杨帆, 马枫, 等. 智能船舶航行功能测试验证的方法体系[J]. 中国舰船研究, 2021, 16(1): 45–50
LIU Jia-lun, YANG Fan, MA Feng, et al. Method system for verification of intelligent ship navigation function test[J]. China Ship Research, 2021, 16(1): 45–50
[3] 高立杰, 位少坤, 汪祥亮. 基于数据驱动的自动驾驶测试场景生成方法[C]//第十八届河南省汽车工程科技学术研讨会论文集, 2021: 236-238.
[4] 周文帅, 朱宇, 赵祥模, 等. 面向高速公路车辆切入场景的自动驾驶测试用例生成方法[J]. 汽车技术, 2021(1): 11–18
ZHOU Wenshuai, ZHU Yu, ZHAO Xiangmo, et al. Automatic driving test case generation method for highway vehicle cut-in scenarios[J]. Automotive Technology, 2021(1): 11–18
[5] 孙崇珊. 基于车路协同混合交通场景要素解析的测试案例生成方法[D]. 北京：北京交通大学, 2021.
[6] 赵祥模, 高赢, 徐志刚, 等. IntelliWay-变耦合模块化智慧高速公路系统一体化架构及测评体系[J/OL]. 中国公路学报, 2022:1-33. http://kns. cnki. net/kcms/detail/61. 1313. U. 20220601. 1805. 004. html
Zhao Xiangmo, Gao Ying, Xu Zhigang, et al. IntelliWay - Variable coupling modular intelligent highway system integration architecture and evaluation system [ J / OL]. China Journal of Highway and Transport http: //kns. cnki. net/ kcms/ detail/ 61. 1313. U. 20220601. 1805. 004.html, 2022:1−33.
[7] 张帅, 邹浩彦, 张磊. 基于模型的飞行管理测试场景自动生成方法[C]//. 2021年中国航空工业技术装备工程协会年会论文集, 2021: 317-321.
[8] 李江坤, 邓伟文, 任秉韬, 等. 基于场景动力学和强化学习的自动驾驶边缘测试场景生成方法[J]. 汽车工程, 2022, 44(7): 976–986
LI Jiangkun, DENG Weiwen, REN Bingtao, et al. Scenario generation method of autonomous driving edge test based on scene dynamics and reinforcement learning[J]. Automotive Engineering, 2022, 44(7): 976–986
[9] 蒋娇. 一种航迹智能控制算法优化研究[D]. 厦门: 集美大学, 2014.
[10] 蒋拯民, 党少博, 李慧云, 等. 自动驾驶汽车场景测试研究进展综述[J]. 汽车技术, 2022(8): 10–22
JIANG Zhengmin, DANG Shaobo, LI Huiyun, et al. Overview of research progress in scenario testing of autonomous vehicles[J]. Automotive technology, 2022(8): 10–22
[11] FENG Gao, et al. A test scenario automatic generation strategy for intelligent driving systems[J]. Mathematical Problems in Engineering, 2019 : 1-10.
[12] W TSENG, C Fan. Systematic scenario test case generation for nuclear safety systems[J]. Information and Software Technology, 2013, 55(2): 344–356
[13] GEMA, PGS B, RCH B, et al. Adaptive generation of challenging scenarios for testing and evaluation of autonomous vehicles[J]. Systems & Software, 2018(137): 197–215
[14] E ROCKLAGE, H KRAFT, A KARATAS, et al. Automated scenario generation for regression testing of autonomous vehicles[C]//2017 IEEE 20th International Conference on Intelligent Transportation Systems (ITSC), 2017: 476-483.
[15] J DUAN, F GAO, YH e, Test scenario generation and optimization technology for intelligent driving systems[C]//In IEEE Intelligent Transportation Systems Magazine, 2022,114(1): 115-127.
[16] S LIU, S NAKAJIMA, Automatic test case and test oracle generation based on functional scenarios in formal specifications for conformance testing[C]// In IEEE Transactions on Software Engineering, 2022, 48(2): 691−712.
[17] SARMIENTO E, LEITE TCSP, ALMENTERO E,et al. Test scenario generation from natural language requirements descriptions based on Petri-Nets[J]. Electronic Notes in Theoretical Computer Science, 2016, 329(329): 123–148
