Personalized teleoperation via intention recognition -

Abstract

One of the major challenges in teleoperation is the recognition of a user’s intended commands, particularly in the manning of highly dynamic systems such as drones. Since the introduction of unmanned aerial vehicles (UAVs), their teleoperation was and still is a challenging task. Significant research was conducted to facilitate and improve UAV maneuverability, and yet many accidents are attributed to human error. The main purpose of this thesis is to build a teleoperation system that allows users to teleoperate any UAV in an intuitive manner. Remote controls (RCs) are commonly used to control UAVs, but with their constant input mapping parameters and fixed configuration, RCs do not suit all users and require significant training. In this work, we are addressing the issue faced with this type of controllers, where multiple new teleoperation techniques are investigated and tested to design a powerful teleoperation system that acts based on the intentions of pilots. Haptic joysticks are usually used as a replacement for RCs in the field of teleoperation. With the conducted experiments, it has been shown that with the appropriate configuration, haptic joysticks are easy to use and help pilots to perform the tasks in an efficient and accurate manner. But studies were mainly focused on using one joystick and only few were interested in the use of multiple joysticks. As the user performance was not improved by using two joysticks, a new approach was adopted to enhance the teleoperation system. We focused on modifying the mapping parameters of the joystick based on the pilot commands. An adaptive gain tuning algorithm is used to map the masters’ inputs (RC inputs) to the slave (UAV) motions to enhance the pilot’s performance, regardless of the subject’s experience. Based on his-her commanding characteristics the input sensitivity and smoothing of the joystick is modified. Experimental results showed a significant improvement when adaptive gain tuning is employed. Once the adaptive gain tuning alg