基于校园场景的无人配送车智能驾驶系统的测试与评价

Test and Evaluation of Intelligent Driving System of Unmanned Delivery Vehicle Based on Campus Scene

摘要: 无人驾驶车辆的测试与评价对保障其行车安全至关重要，本文对校园环境下的无人配送车展开研究，对其智能驾驶系统进行测试分析与评价。通过校园场景的解构，运用PICT (Pairwise Independent Combinatorial Testing tool)组合测试工具生成场景重构组合集，并建立树状模型，分析校园测试场景库；改进Pairwise算法，在满足覆盖度的前提下，运用概率决策使生成的测试用例集数量明显降低，对这组矛盾问题进行有效控制；选取校园典型道路和天气类型，合理规划出无人车配送路线，并结合实地考察合理设计测试场景；基于层次分析法确定权重，运用模糊综合评价法对无人配送车进行综合评价。

Abstract: The test and evaluation of unmanned vehicles is very important to ensure their driving safety. This paper studies the unmanned delivery vehicles in the campus environment, and tests, analyzes and evaluates its intelligent driving system. Through the deconstruction of campus scenes, the Pairwise Independent Combinatorial Testing tool PICT (Pairwise Independent Combinatorial Testing tool) was used to generate scene reconstruction Combinatorial sets, and the tree model was established to analyze the campus test scene library. The Pairwise algorithm was improved to reduce the num-ber of test case sets generated by probabilistic decision making under the premise of satisfying the coverage degree, and the contradictory problems were effectively controlled. The typical roads and weather types on campus were selected to reasonably plan the unmanned vehicle distribution routes, and the test scenarios were reasonably designed based on field investigation. Based on the analytic hierarchy process, the weight is determined, and the fuzzy comprehensive evaluation method is used to comprehensively evaluate the unmanned delivery vehicle.