Augmented Reality for Teleoperated Human-Robots interaction

Nowadays, robots are becoming more and more complex, providing increasingly functionalities, precision and effectiveness. Yet, the amount and variety of tasks that robots can accomplish autonomously is limited. For this reason, when dealing with complex tasks, usually robots are controlled by a human operator through teleoperation. Teleoperation allows humans to perform tasks in environments that are harsh or not accessible and to reduce travel needs as in the case of telemedicine. The downside of teleoperation comes from a reduced perceptual capability and the control of a robot that has a different kinematic structure or capabilities than the human, requiring specific training. To overcome these difficulties it is possible to employ Mixed and Augmented Reality display and feedback. The idea is to overlay on the visual feedback received from the remote robot virtual cues providing information about the remote environment or about the task in execution, which are not directly inferable from the video. Virtually generated features can be used not only during the actual task execution, but also for training purposes. In order to recreate a specific scenarios scientist and engineers started to create Virtual Environments (VE) that are replicas of real world scenarios. In this case the operator controls a simulated version of the target robot. This thesis describes a set of software approaches and design solutions finalized at the creation of integrated real-time AR/VR applications, with a focus on fast prototyping and reconfigurability. The first part of the thesis presents the state of the art software solutions created to provide tools that ease the development of AR/VR applications. The software is described and presented along with real examples. The second part, instead, focuses on the study and analysis of the impact of AR and VR in two specific scenarios: industry and medicine.