Shared Control and Authority Distribution for Robotic Teleoperation

Human robot interaction is at the core of the robotic developments over the years. In fact, while robotic autonomy has been perfected in structured and predictable environments, the human input is still needed in many cases, especially in a variable setting. This thesis falls under the general human robot interaction topic. More specifically, it deals with remote interaction under the form of teleoperation. We design novel shared control architectures for safe and intuitive robotic teleoperation, with various types of interaction and different levels of automation. We focus in this work on the use of haptics as a means of interaction, whether for providing contact forces as a feedback to the user, or as a means of guiding the user in the task execution. Starting with a robotic grasping and manipulation application, we design new haptic shared control targeting both the pre-grasp and the post-grasp phases. One shared control architecture aims to minimize the robot torques in a post-grasp phase, while the other focuses on minimizing the human workload during task execution. We also target a robotic cutting application, and design a novel haptic shared control technique for robotic cutting imposing non-holonomic constraints motivated by the cutting task. We finally highlight the role of haptic interaction and human input in the design of autonomous grasping algorithms, using the human experience and haptic feedback to learn autonomous grasping policies for compliant objects.