Cooperative interaction with cyclists in automated driving – KIRa

开姆尼茨工业大学（Chemnitz University of Technology）的跨学科项目KIRa旨在研究自动驾驶车辆与弱势道路使用者（VRU）之间合作互动的解决方案，重点是在障碍物情况下骑自行车的人和行人。研究目标将通过工程与交通心理学的紧密跨学科合作来实现。主要场景是低速城市地区（如停车场）复杂的交互情况，需要隐式和/或显式通信。该项目是SSP 1835提案中“3.2 Situationsprädiktion und Intentionserkennung”和“3.6系统经济学”研究领域的一部分，也是项目前KIVI的一贯延续。KIVI的主要目标和扩展是检查自动驾驶车辆如何评估和预测与自行车和受阻行人的合作情况，可以使用哪些合作信号，如何将这些信号参数化，以及如何有效地实施连接的交互架构。KIRa结合了跨学科的方法和创新的研究方法，如视频模拟研究，结合具身认知，vru数据的分散获取和互联条件下的信息交换。工作计划从在预先指定的具有交互需求的真实世界场景中使用不同传感器获取视频和测量数据开始。这些场景将被整合到spp1835现有的用例目录中。这些轨迹将使用SPP中定义的xml格式与所有SPP项目共享。基于视频材料，进行实验性用户研究，以确定有关隐式和显式通信的具体交互参数。算法将根据所有道路使用者的运动学和地理参考数据，结合基础设施传感器信息开发。这些算法的目的是预测VRU的运动轨迹，并对VRU的行为模式进行基于情境的评估，以参数化自动测试车辆的机动规划。基于所有结果，将进行全面的道路测试，包括行人，骑自行车者和自动测试车辆。这些真实世界的测试允许测试和评估所有开发的算法、分散数据采集、vru轨迹的通信和预测。

The interdisciplinary project KIRa at Chemnitz University of Technology aims at investigating solutions for cooperative interaction between automated vehicles and vulnerable road users (VRU) with a focus on bicyclists and pedestrians in obstruction situations. The research goals will be realized through a close interdisciplinary cooperation between engineering and traffic psychology. Main scenarios are complex interaction situations in low-speed urban areas (such as parking areas) which require implicit and/or explicit communication. The project is part of the research areas "3.2 Situationsprädiktion und Intentionserkennung" and "3.6 Systemergonomie" of the SSP 1835 proposal and a consistent continuation of the pre-project KIVI. Main goals and extensions of KIVI are the examination of how automated vehicles can assess and predict cooperation situations with bicycles and obstructed pedestrians, which cooperation signals can be used, how these signals need to be parameterised and how connected interaction architectures can be implemented efficiently. KIRa combines interdisciplinary approaches and innovative research methods such as video simulation studies incorporating embodied cognition, decentral acquisition of VRU-data and information exchange under connected conditions. The work plan starts with the acquisition of video- and measurement data using different sensors in pre-specified real-world scenarios with interaction demands. These scenarios will be integrated in the existing use-case-catalogue of the SPP 1835. The trajectories will be shared with all SPP-projects using the XML-format defined in the SPP. Based on the video material, experimental user studies are carried out to determine specific interaction parameters regarding implicit and explicit communication. Algorithms will be developed based on kinematic and georeferenced data of all road users in combination with infrastructure sensor information. Aim of the algorithms is the prediction of VRU-trajectories and the situation-dependent evaluation of VRU behaviour patterns in order to parameterise manoeuvre planning of the automated test vehicle. Based on all results, comprehensive on-road tests will be carried out including pedestrians, bicyclists and the automated test vehicle. These real-world tests allow for testing and evaluating all developed algorithms, decentral data acquisition, communication and prediction of VRU-trajectories.