A Semantics of Spatial Expressions for interacting with unmanned aerial vehicles

A Semantics of Spatial Expressions for interacting with unmanned aerial vehicles Lucas C. F. Domingos Centro Universit´ario FEI, SBC, S˜ao Paulo, Brazil ldomingos@fei.edu.br Paulo E. Santos College of Science and Engineering Flinders University Adelaide - South Australia paulo.santos@flinders.edu.au Abstract This paper describes an investigation of estab- lishing communication between a quadrotor and a human through qualitative spatial rela- tions allied with an off-the-shelf speech recog- nition software. The quadrotor used in this research was equipped with GPS, IMU sen- sors, and radio communication, which was connected to a computer acting as a ground station. The ground station was implemented to interpret the received commands, correctly providing answers to the user according to an underlying qualitative reasoning formalism. The results obtained during the tests show that the error rate related to the answers given by this system was less than five per cent for ver- tical and radial dimensions. In contrast, com- mands related to the horizontal extent had an error rate of almost ten per cent. 1 Introduction Unmanned aerial vehicles (UAV) have been gain- ing popularity in recent years due to their potential for novel applications (Shakhatreh et al., 2019). One of the most well-known types of UAVs is the quadrotor, owing to their fair cost-benefit and a large number of off-the-shelf programming tools available for application development. Some po- tential current and future activities involving UAVs include, for instance, mapping large areas (Achte- lik et al., 2009), recording movie scenes (Fleureau et al., 2016), and search and rescue missions (Mal- faz and Salichs, 2004). One of the challenges for achieving a large-scale use of UAV applications, however, comes from the need to make more natural the way humans interact with such vehicles, especially for the non- specialised public (Franchi et al., 2012). This issue justified the development of an area of research known as human-robot interaction (HRI). HRI aims to develop strategies for facilitating the interaction with robotic agents in various situations, such as teaching children, rehabilitation, housework and many others. However, HRI is still to be considered in the context of UAV applications. Nevertheless, the most direct way to achieve a high level of com- munication and understanding between robots and humans is the vocal commands usage to transmit and answer the commands between these agents. When two people want to talk to each other in everyday situations, they rarely use quantitative in- formation, especially when talking about space and its relations (Aoyama and Shimomura, 2005). For example, when we say to a child to catch something at a table we do not tell the distance in meters or the relative altitude, we just give the basic qualitative information, like if it’s close or far, under the table or on it. This observation motivated the present in- vestigation, which aim is to develop new methods of human-robot communication using qualitative information. In general terms, this work aims to bridge the gap in the communication between a human and a quadrotor using speech recognition and a qualitative way …

Content truncated to 500 words