Safe Adaptive Social Cyber Physical Systems

社会协作机器人（cobots）是一种新兴技术。协作机器人作为一个团队与人类工人一起完成复杂的任务，与手工劳动或机器人系统相比，提高了活动的生产率。这种系统的部署在很大程度上依赖于保证操作安全的软件开发技术。其中一种技术称为基于模型的软件开发（MBSD）。然而，随着机器学习的兴起和机器人系统适应性的增强，软件安全现在变得更加复杂。******本提案涉及几种技术的应用，以保证协作机器人系统的安全性，同时保持其适应性。在这些技术中，我们使用了使用离散事件系统和模型构建约束的模型的可解释性，不确定性下的机器人推理，使用生物信号估计协作任务中的人类认知负荷，以及通过使用MBSD进行实时保证。******要求的设备包括两个协作机器人和一个脑电图（EEG）结合功能近红外光谱（fNIRS）耳机。该设备将与皇后大学已有的相机测量系统一起用于实验室空间。摄像机系统将用于跟踪机器人和人类参与者的运动。协作机器人将被编程为独立行动，并在装配任务中与人类合作。由于人类的运动不是完全确定的，机器人在有效解决复杂任务的同时必须处理不确定性。******该设备将用于女王大学的四个研究项目：三个在计算机学院，一个在电气和计算机工程系，最初将涉及20多名高素质人员。到2021年，将有近30个HQP直接参与该设备的研究。重点将放在招收多样化的学生群体，这将从实验室已经提供的多样化研究环境中受益。随着全球化市场对竞争优势的需求增加，我们的培训计划非常适合帮助满足加拿大企业和研究机构在机器学习和机器人技术方面的需求，以改善工业任务的结果。

Social collaborative robots, known as cobots, are a new emerging technology. Cobots work with human workers as a team in complex tasks, increasing the productivity of the activity when compared to only manual labour or robotic systems. The deployment of such systems is largely dependent on software development techniques that guarantee the safety of operation. One such technique is known as Model-based Software Development (MBSD). However, with the rise of machine learning and increased adaptability of robotic systems, software safety is now more complex to achieve.******This proposal deals with the application of several techniques that guarantee the safety of cobot systems while maintaining their adaptability. Among these techniques, we use the interpretability of models using discrete event systems and model construction constraints, robotic reasoning under uncertainty, estimation of human cognitive load in cooperative tasks using biological signals, and real-time guarantees through the use of MBSD. ******The equipment requested consists of two collaborative robots and an electroencephalogram (EEG) combined with functional near-infrared spectroscopy (fNIRS) headset. The equipment will be used in a lab space with a camera measurement system already available at Queen's University. The camera system will be used to track the motion of the robot and human participants. The collaborative robots will be programmed to act independently and collaborate with humans in assembly tasks. Since human motion is not completely deterministic, robots will have to deal with uncertainty while efficiently solving complex tasks.******The equipment will be used by four research programs at Queen's University: three at the School of Computing and one at the Department of Electrical and Computer Engineering and will involve initially over 20 highly qualified people. By 2021, close to 30 HQP will be directly involved in the research with the equipment. Emphasis will be given to the recruitment of a diverse student population that will benefit from the diverse research environment already available at the labs. With the increased need of competitive advantages required in the current globalized marketplace, our training programs are well suited to help fill the need of Canadian businesses and research institutes in machine learning and robotics applied to improving the outcome of industrial tasks.