Multiple-layered human interface for safety control and sharing of situation awareness in dynamical environments with risks

本课题研究了在动态变化的时间环境下支持人机协同的人机界面设计。可能在情况识别、决策以及行动选择和实施方面受到限制。功能分配需要是动态的和情况适应的，以适当地支持人类。因此，机器被赋予了各种类型的智能。智能机器现在可以感知和分析情况，决定必须做什么，并实施控制行动。然而，使用这类智能机器工作的人类确实经常遭受自动化的负面后果，例如环路外的性能问题、丧失情况意识、自满或过度信任，以及自动化引发的意外。本研究以自适应巡航控制(ACC)系统为例，提出了多层人机界面的概念，将信息提供给驾驶员，以便驾驶员与自动化系统共享态势识别和意图，并在紧急情况下动态交换权限，以确保系统安全。通过对不同类型场景的一系列认知实验，对人机界面的有效性进行了分析和评估。此外，已经开发了计算机模拟方法来调查驾驶员对自动化的过度信任导致的安全降级程度，在这种情况下传统的认知实验方法可能不适用或不可行。已经证明，确保系统安全的人和自动化之间的功能分配的有效策略可能属于不满足传统的以人为中心的自动化所假定的条件的类别。

This research project has investigated design of human interface to support human-machine collaborations under dynamically changing environment in which allotted time. may be limited for situation recognition, decision making, and action selection and implementation. Function allocation needs to be dynamic and situation-adaptive to support humans appropriately. Machines have thus been given various types of intelligence. Intelligent machines can now sense and analyze situations, decide what must be done, and implement control actions. It is true, however, humans working with such smart machines often suffer negative consequences of automation, such as the out-of-the-loop performance problem, loss of situation awareness, complacency or over-trust, and automation-induced surprises. By taking the adaptive cruise control (ACC) system as a real-world example of adaptive systems, this research project has developed the concept of multi-layered human interface in which information is provided with human drivers so that they can share situation recognition and intentions with automated systems and authority may be traded dynamically in an emergency to assure system safety. The efficacy of the human interface has been analyzed and evaluated through series of cognitive experiments with various types of scenarios. Furthermore, computer simulation methods have been developed to investigate degree of safety degradation due to drivers' over-trust in automation, in which conventional cognitive experimental approach may not be applicable or feasible. It has been proven that an effective strategy for function allocation between humans and automation assuring system safety can lie within the category that does not fulfill conditions assumed for a conventional human-centered automation.