自动驾驶路测融合度建模与分析

Integration Degree Modeling and Analysis for Autonomous Vehicles in Road Testing

摘要:

选取5个指标构建评价指标体系，然后采用皮尔生长曲线函数和负指数函数对正向指标和逆向指标进行标准化，并利用专家打分和层次分析法量化各指标权重，最后建立自动驾驶路测融合度模型。对基于上海市2类城市道路场景和1类高速公路场景的自动驾驶路测实测数据与仿真结果开展案例分析，验证路测融合度模型的合理性和有效性。结果表明：无论对于城市道路场景还是对于高速公路场景，避险脱离率是表征自动驾驶车融入到现有道路交通系统程度的最重要指标；高速公路场景下的自动驾驶路测融合度显著高于城市道路场景下的自动驾驶路测融合度；加大自动驾驶测试里程、测试时长和测试场景复杂度，可以促进自动驾驶技术的成熟。

Abstract:

An evaluation framework based on five evaluation indicators was proposed. The Pearl growth curve function and negative exponential function were selected to standardize the positive and negative indicators， respectively. Then， the expert scoring method and analytic hierarchy process were used to determine the weight of each indicator. Finally， the integration degree model of autonomous driving road testing was built. The case study， based on the measured data and simulation results of automatic driving road testing under two types of urban road scenarios and one type of highway scenario in Shanghai， is conducted to verify the validity and effectiveness of the integration degree model. The results show that the risk-avoiding disengagement frequency is the most critical evaluation indicator which characterizes the integration degree of autonomous vehicles into existing road traffic system under both urban road scenarios and highway scenario； the integration degree under the highway scenario is significantly higher than that under the urban road scenario； increasing the testing mileage， the testing duration and the complexity of testing scenarios can promote the maturity of autonomous vehicle technology.