An interaction method for search & rescue aerobots
This paper is interested in how the increasing leagues of human operators interact with small multi-rotor UAVs. The multimodal speech and visual gesture (mSVG) technique is basically the multimodal combination of speech and visual gesture, a method that leverages familiar human-human type interaction, in human aerobotic interaction. Speech was captured via a microphone and processed to recognise set control words. Visual gesture was captured via a camera connected to the aerobot SBC computer. The mSVG control navigation was simulated on a rotors gazebo firefly UAV simulator in an open world environment. In each case, a series of command such as ‘go forward’, ‘go up half metre’, ‘go right one metre’, ‘hover at three metre’, ‘and’, ‘hover’, ‘go forward backward two half metre’, ‘patrol’, etc. were successfully tested. The main limitations of the proposed system due to external noise and poor visibility are discussed.
Abioye, Ayodeji, Opeyemi, Prior, Stephen, Thomas, Trevor, Saddington, Peter and Ramchurn, Sarvapali (2017) A practical mSVG interaction method for patrol, search, and rescue aerobots. UK-RAS Conference on ‘Robotics and Autonomous Systems’ (RAS 2017): Robots working for and and among us, Bristol, United Kingdom. 12 Dec 2017. Read publication.