Improving Robotic Teleoperation through Human-Robot Interaction and Robotic Autonomy

遥操作的总体目标是让操作员对远程环境有一种情景感知(即身临其境的感觉)，并能够方便地交互。核电站中远程操纵器的操作员执行检查和维护任务，包括复杂的操作。几项研究表明，环境的非结构化性质、任务的复杂性以及核工业的保守性导致远程操作员更愿意控制并随时充分了解多个系统的当前和未来行为，从而限制了自主远程操作或机器人的部署。因此，这些机械系统的操作员同时与几种媒介交互，例如各种显示器、机器人远程操纵系统、信息系统和其他同事。这些交互作用给操作员带来了巨大的信息处理水平，导致了高认知工作量，从而影响了遥操作的性能和安全。本博士研究针对遥操作中的自动化挑战，专注于提高操作员的性能和安全性。它承认，在复杂的非结构化环境中，操作员更倾向于控制，这阻碍了自主系统的部署。它建议将自适应机器人辅助和响应界面相结合，以动态响应操作员的脑力工作量(MWL)需求、任务复杂性和结构。为了实现这一目标，将达到以下目标：确定遥操作的客观性能和脑力负载指标；使用任务分析确定体力和脑力负载的来源；开发评估遥操作中性能和脑力负载的方法；基于体力和精神需求设计机器人辅助方案，以指导操作者完成遥操作任务。几篇文献已经演示了基于触觉的触觉引导如何应用于机器人辅助。例如，触觉共享控制和触觉交易控制，操作者和系统分别共享和轮流执行控制任务。这项工作将探索如何部署这些援助计划，以响应操作员在结构化环境中对重复和可预测任务的身体和认知需求。它将结合一些生理、主观和性能指标来改进对操作员脑力劳动的评估和评价，以了解其与遥操作任务的关系。有了这些改进的系统，操作员可以更方便、更快、更准确地执行他们的工作，从而提高安全性并降低运营成本。

Teleoperation's overall goal is to give the operator situational awareness (i.e., a sense of being there) of the remote environment and the ability to interact conveniently. Operators of telemanipulators in nuclear plants perform inspections and maintenance tasks, including complex manipulations. Several studies reveal that the unstructured nature of the environment, task complexity, and the conservative nature of the nuclear industry cause teleoperators to prefer to be in control and fully aware of the multiple systems' current and future behaviours at all times, limiting the deployment autonomous teleoperation or robotics. As a result, the operators of these mechanical systems interact with several mediums, such as various displays, robotic telemanipulation systems, information systems, and other colleagues, at once. These interactions impose overwhelming levels of information processing on the operator, leading to a high cognitive workload, which impacts performance and safety in teleoperation.This PhD research addresses the automation challenge in teleoperation, focusing on enhancing operator performance and safety. It acknowledges the preference for operator control in complex, unstructured environments, hindering the deployment of autonomous systems. It proposes integrating adaptive robotic assistance and responsive interfaces that dynamically respond to an operator's mental workload (MWL) demands, task complexity and structure. The following objectives will be met to achieve this aim:To identify the objective performance and mental workload metrics for teleoperation.To determine the sources of physical and mental workload using task analysis.To develop methods for assessment of performance and mental and physical workloads in teleoperation.To design robotic assistance schemes based on physical and mental demands to guide operators through teleoperation task completions.Several pieces of literature have demonstrated how haptic guidance based on the sense of touch can be deployed for robotic assistance. For example, haptic shared control and haptic traded control, which the operators and the system share and take turns in the control task, respectively. This work will explore how these assistance schemes could be deployed to respond to an operator's physical and cognitive demands within structured environments for repetitive and predictable tasks. It will combine some physiological, subjective and performance measures to give an improved assessment and evaluation of the operator's mental workload to understand its relationships with teleoperation tasks. With these improved systems, operators can perform their work more conveniently, faster, and accurately, increasing safety and reducing operating costs.