Let’s Fly! An Analysis of Flying FPV Drones Through an Online Survey Dante Tezza Denis Laesker Abstract University of South Florida University of South Florida First-person view (FPV) drones provide an immersive flight Tampa, FL 33620, USA Tampa, FL 33620, USA experience for pilots. In FPV flying, the pilot wears a pair of dtezza@mail.usf.edu dlaesker@mail.usf.edu goggles that display the video feed from the drone in real- time. This allows them to fly the drone as if they were sitting on top of it, thus creating an immersive experience simi- lar to virtual reality and giving the sensation of free flight. Due to these characteristics, FPV flying is becoming pop- ular for recreational purposes (e.g. drone racing). In this Derek Caprio Marvin Andujar study, we conducted an online survey with 515 FPV pilots University of South Florida University of South Florida to explore their preferences and give human-drone inter- Tampa, FL 33620, USA Tampa, FL 33620, USA action researchers an understanding of the FPV commu- derekc1@mail.usf.edu andujar1@usf.edu nity and how pilots interact with the drones. In this paper, we present that the majority of pilots prefer acrobatic flight mode for both racing and freestyle flying. Additionally, we found that FPV flying introduces users to technical fields as the majority of the pilots build their drones, even without having any previous technical background. Lastly, we also present how pilots prefer to interact with remote controllers. Author Keywords This paper is published under the Creative Commons Attribution 4.0 International Drones, drone racing, first-person view, human-drone inter- (CC-BY 4.0) license. Authors reserve their rights to disseminate the work on their personal and corporate Web sites with the appropriate attribution. action, human-robot interaction Interdisciplinary Workshop on Human-Drone Interaction (iHDI 2020) CHI ’20 Extended Abstracts, 26 April 2020, Honolulu, HI, US © Creative Commons CC-BY 4.0 License. CCS Concepts •Human-centered computing → Human computer inter- action (HCI); User studies; Introduction Human-drone interaction research (HDI) can be defined as Oftentimes we hear people express their desire to fly; the the study focused on understanding, designing, and eval- idea of being able to see and explore the world from the uating drone systems for use by or with human users[14]. skies has fascinated humans for centuries. The Wright Current human-drone interaction research has focused on brothers achieved the first successful controlled flight in developing natural interaction[5], flying user interfaces [7] 1903 [15], and since then, aviation has been evolving and new control modalities(e.g. brain-controlled drones[11]), becoming more and more ubiquitous in society. These de- social drones[3], and even using drones to provide haptic velopments have led to modern unmanned areal vehicles feedback for virtual reality environments[9]. Drones that capable of flying autonomously or being remote-controlled, broadcast images in real-time have been studied in appli- these aircraft are commonly referred to as drones. They are cations like search and rescue and disaster relief [8]. How- used in a broad range of applications such as photography, ever, there is a lack of research on the community of pilots Figure 1: FPV goggles. natural disaster response [2], agriculture [12], and drone who fly FPV recreationally, most likely due to the recency racing [4]. Drone usage is increasing and is expected to of the field. The ability to fly a drone with the sensation continue to do so, the Federal Aviation Administration (FAA) of being onboard allows the creation of new user experi- projects that 3.8 million drones will be registered in their ences, interactions, sports, and novelty drone use cases. database in the United States alone by 2022 [6]. However, Therefore, we believe that human-drone interaction (HDI) drones are remotely operated by a human on the ground research could benefit from studying the user interaction and generally lack the ability to provide an immersive expe- with FPV drones. rience, which is an important aspect in the remote operation of robots[1]. In this study, we evaluated the FPV drone community and their flight preferences through an online survey completed Recently, a new type of flying drones, known as first person by 515 FPV pilots. We surveyed FPV pilots to better un- view (FPV), has emerged providing users a more immer- derstand their user experience when flying from a human- sive experience while flying. In this modality, the drone is drone interaction perspective. Our contributions in this pa- equipped with a front-facing camera which transmits live per are the results of this survey, allowing HDI researchers Figure 2: FPV pilot controlling a video to a pair of googles worn by the pilot. This gives the to evaluate how pilots interact with FPV drones and better drone. pilot the impression that they are sitting on top of the drone understand their user experience. In this survey, we found and leads to an immersive experience similar to virtual re- that the majority of pilots prefer acrobatic flight mode for ality. FPV flying gives the pilot the sensation of free flight, both racing and freestyle flying. Additionally, we found that making it especially popular among drone pilots who enjoy FPV flying introduces new users to technical fields as the flying recreationally. Acrobatic flights (known as freestyle) majority of the pilots build their drones, even without having and racing competitions are common within the FPV com- any previous technical background. Lastly, we also present munity. An example of FPV goggles can be seen in Figure how pilots prefer to interact with remote controllers. 1, followed by a pilot flying FPV in Figure 2, and a FPV im- age (displayed in goggles) in Figure 3. Figure 3: FPV image displayed in goggles. Study Design & Methodology Results & Discussion To understand the FPV drone community and the experi- FPV Racing and Freestyle ence of flying FPV drones, we conducted an anonymous First-person view flying is divided into two major categories online survey with 51 questions with 515 FPV pilots. The of flying styles: racing and freestyle. Drone racing is an survey was administered through Qualtrics and data was emerging and competitive sport in which pilots fly FPV collected over a period of four months. Participants were drones in complex 3D courses against each other, aiming to required to be at least 18 years old and have experience be the fastest pilot on the track [4]. Drone racing started as flying FPV drones. Before completing the survey, partici- an amateur sport in Australia during the year of 2014 and pants had to digitally sign an informed consent form. Re- grew in popularity due to pilots posting racing videos in so- cruitment was performed solely online and shared on Face- cial media. Drone racing is significantly more complex than book , Twitch , Discord , Twitter, and Reddit on groups and flying non-FPV drones, it requires long practice periods and channels related to FPV drones. The survey collected data high level of skills [13]. Freestyle flying is a broader con- regarding (1) pilots’ backgrounds (e.g. electronics knowl- cept, as there are no specific rules or competitions for this edge) and how they impact their current flight experiences, category. There is no previous formal definition of freestyle (2) pilots’ flight preferences (e.g. preferred flight modes, re- flying, therefore, we derive its definition from another ex- mote controller grips), (3) pilots’ preferences towards equip- treme activity, freestyle BMX; where its pilots spend their ment (software and hardware), and (4) how pilots learned time performing tricks and stunts rather than competing in how to fly FPV. races[10]. Analogously, we define freestyle flying as the category where pilots fly FPV drones to explore spaces, Participants perform tricks and stunts. A total of 515 FPV pilots completed the survey. Of these 515 participants, 505 (98.06%) participants are male, 5 Our analysis of the 515 FPV pilots show that 43.08% of (0.97%) are female, and 5 (0.97%) do not identify as either them fly freestyle only, 8.33%, fly only for racing purposes, male or female. Additionally, 79 (15.34%) are between the and 48.57% fly both racing and freestyle. Also, 80% of rac- ages of 18 and 24, 133 (25.83%) are between the ages of ing pilots compete at some level, compared to only 13% of 25 and 34, 176 (34.17%) are between the ages of 35 and freestyle pilots. This is attributable to the more competitive 44, 87 (16.89%) are between the ages of 45 and 54, 34 nature of racing sports. Although there are official FPV rac- (6.60%) are between the ages of 55 and 64, and 6 (1.17%) ing leagues which host competitions, no such leagues exist are 65 years of age or older. Of these same participants, for freestyle pilots. This is also a plausible explanation for 27 (5.24%) are Hispanic or Latino, 439 (85.24%) are Cau- why 15% of racing pilots declared to receive some sort of casian, 2 (0.39%) are African-American, 2 (0.39%) are sponsorship, as these competitions are often televised and Black, 13 (2.52%) are Asian, and 32 (6.21%) are something draw public attention. In contrast, only 3% of freestyle pilots other than the ethnicity listed above. receive sponsorship. We expect that as FPV sports con- tinue to grow in popularity, official freestyle competitions will emerge and it will become more common for freestyle pilots to receive sponsorship. Flight Mode % of Pilots FPV Flight Modes that the attitude commanded by the pilot is performed cor- Position Hold 0.81% The characteristics on how a FPV drone responds to pi- rectly. These characteristics allow racers to better maneu- Angle 6.10% lots input is dependent on the concept of flight modes. This ver around obstacles on the track, as well as freestylists Acrobatic 92.07% setting can be set on the flight controller and dictates how to perform stunts that would not be possible in other flight Unknown 0.41% the pilot can interact with the drone. Non-FPV drones are modes. Other 0.61% capable of many flight modes, including assisted and au- tonomous modes. However, FPV flying aims to provide the FPV Remote Controllers Controller Grip % of Pilots The remote controller (RC) acts as an interface between pilot with the most control authority over the drone, there- Thumb 53.46% fore there are only two main flight modes that are com- the pilot and the drone, therefore it is important to under- Pinch 20.9.% monly used: stand pilots’ preferences when interacting with such de- Hybrid 24.80% vices. There are two factors that influence how FPV pilots Unknown 0.81% interact with the RC itself: the form of grip, and RC mode. • Angle/Stabilized - This is an auto-stabilization mode There are three main forms of grips in which the pilot holds RC Mode % of Pilots in which the pilot input command is translated di- the controller: thumbs, pinch, and hybrid. These are dis- Mode 1 6.10% rectly to aircraft attitude, and the aircraft stabilizes played in Figures 4 5 6 respectively. Table 1 also shows Mode 2 87.40% itself if the pilot does not send any command. In other that the majority of pilots (53.46%) prefer to hold their con- Mode 3 0.81% words, the movement in the remote controller sticks trollers using the "thumb grip", followed by hybrid(24.80%) Mode 4 1.02% is translated to the angle in which the drone flies, and and pinch (20.9%). However, hybrid and pinch are similar Unknown 4.67% the drone levels itself when the sticks are centered in grips, and if they are analyzed together, the gap between the controller. their use and thumb grip is not as large. The form of grip Table 1: FPV pilots’ preferences. can influence the pilot’s interaction with the drone in terms • Acro/Rate - In this mode, the aircraft does not sta- of control latency, accuracy, and comfort. Future studies bilize itself, and the stick movement is translated to could objectively evaluate how each grip mode impacts the angular velocity to its correspondent axis. The move- human-drone interaction. ment in the remote controller stick dictates how fast the drone spins on each axis, and in the absence of There are also four RC modes that dictate how the RC gim- input, the drone will maintain the current angle in- bal sticks are translated to drone commands, as shown in stead of stabilizing itself. Figure 7. Our results (see Table 1) shows that a large ma- jority of pilots (%87.40) prefer to fly in Mode 2, where the As shown in Table 1, 474 out of 515 (92%) pilots surveyed left gimbal translates to throttle and yaw commands, while selected acro as their main flight mode. Although acro flying the right gimbal translates to pitch and roll commands. has a steep learning curve, it provides the highest degree STEM Skills Obtained through FPV Drones of freedom for pilots to control their drone. This mode gives FPV drones can be an effective and highly engaging method the pilot full control over the quad-copter attitude and makes to introduce STEM education as most FPV drone pilots are the drone behavior predictable as the flight controller does also involved in building their systems. In fact, 93.40% of not try to adjust the drone attitude. Instead, it only ensures FPV pilots surveyed have built their own drone before. More importantly, despite 147 pilots (28.54% of all participants) declaring not having any prior electronics background, all of them have now built at least one drone and 90 (61.22%) out of the 147 have built 5 or more drones. STEM skills which are developed through building FPV drones include both hardware (e.g. soldering, electronics) and software skills (e.g. compiling software and flashing microprocessors). Other skills can also be learned such as flight dynamics and tuning for multi-rotor aerial vehicles, as well as 3D print- ing. Since the main flight controller software projects are (a) Mode 1 (b) Mode 2 open source, they can also be used to learn software engi- Figure 4: Thumbs grip neering topics. Finally, when pilots were asked to give open tips for beginners, several of them suggested beginners build their own drone so that they know how to fix it when it becomes damaged after eventual crashes. The above results demonstrate that as users start FPV fly- ing for recreational purposes, they also get introduced to STEM topics. Therefore, we suggest that FPV drone pro- grams (e.g. summer camps, demonstrations, events) can (c) Mode 3 (d) Mode4 be used to introduce new students to STEM. Our results also demonstrate a small female presence in the FPV com- Figure 7: Types of remote controller modes. munity ( 1% of participants). Therefore, we suggest FPV drone programs dedicated to female participants. Increas- ceive sponsorships as opposed to freestyle pilots. We can ing the number of female FPV pilots could be an effective also see that the acrobatic flight mode and RC mode 2 are Figure 5: Pinch grip approach to increase STEM diversity. strongly preferred above all other flight modalities, and that the majority of pilots prefer to hold their controllers using the Conclusion "thumb grip." Furthermore, pilots’ survey responses show In summary, FPV flying is growing in popularity as a way to that the FPV hobby has taught them STEM skills that they provide a more immersive experience to drone pilots, es- did not have previously. pecially for those who fly recreationally. At the same time, this topic lacks the attention from the HDI research com- REFERENCES munity. To better understand the interaction between FPV [1] Sigurdur O Adalgeirsson and Cynthia Breazeal. 2010. pilots and drones, we conducted an online survey with 515 MeBot: a robotic platform for socially embodied FPV pilots. Responses from the survey indicate that pi- presence. In Proceedings of the 5th ACM/IEEE lots who fly for racing are more likely to compete and re- international conference on Human-robot interaction. Figure 6: Hybrid grip. IEEE Press, 15–22. [9] Pascal Knierim, Thomas Kosch, Valentin Schwind, Markus Funk, Francisco Kiss, Stefan Schneegass, and [2] Ludovic Apvrille, Tullio Tanzi, and Jean-Luc Dugelay. Niels Henze. 2017. Tactile drones-providing immersive 2014. Autonomous drones for assisting rescue tactile feedback in virtual reality through quadcopters. services within the context of natural disasters. In 2014 In Proceedings of the 2017 CHI Conference Extended XXXIth URSI General Assembly and Scientific Abstracts on Human Factors in Computing Systems. Symposium (URSI GASS). IEEE, 1–4. ACM, 433–436. [3] Dante Arroyo, Cesar Lucho, Silvia Julissa Roncal, and [10] Wade Gordon James Nelson. 2006. Reading cycles: Francisco Cuellar. 2014. Daedalus: a sUAV for The culture of BMX freestyle. Ph.D. Dissertation. human-robot interaction. In Proceedings of the 2014 McGill University. ACM/IEEE international conference on Human-robot interaction. ACM, 116–117. [11] Amin Nourmohammadi, Mohammad Jafari, and Thorsten O Zander. 2018. A survey on unmanned [4] Amirreza Barin, Igor Dolgov, and Zachary O Toups. aerial vehicle remote control using brain–computer 2017. Understanding Dangerous Play: A Grounded interface. IEEE Transactions on Human-Machine Theory Analysis of High-Performance Drone Racing Systems 48, 4 (2018), 337–348. Crashes. In Proceedings of the Annual Symposium on Computer-Human Interaction in Play. ACM, 485–496. [12] Vikram Puri, Anand Nayyar, and Linesh Raja. 2017. Agriculture drones: A modern breakthrough in [5] Jessica R Cauchard, Kevin Y Zhai, James A Landay, precision agriculture. Journal of Statistics and and others. 2015. Drone & me: an exploration into Management Systems 20, 4 (2017), 507–518. natural human-drone interaction. In Proceedings of the 2015 ACM international joint conference on pervasive [13] Roberto Ribeiro, João Ramos, David Safadinho, and and ubiquitous computing. ACM, 361–365. António Manuel de Jesus Pereira. 2018. UAV for [6] FAA. Unmanned Aircraft Systems. (????). https: Everyone: An Intuitive Control Alternative for Drone //www.faa.gov/data_research/aviation/aerospace_ Racing Competitions. In 2018 2nd International forecasts/media/Unmanned_Aircraft_Systems.pdf Conference on Technology and Innovation in Sports, Health and Wellbeing (TISHW). IEEE, 1–8. [7] Markus Funk. 2018. Human-drone interaction: let’s get ready for flying user interfaces! Interactions 25, 3 [14] Dante Tezza and Marvin Andujar. 2019. The (2018), 78–81. State-of-the-Art of Human–Drone Interaction: A Survey. IEEE Access 7 (2019), 167438–167454. [8] Michael A Goodrich, Bryan S Morse, Damon Gerhardt, Joseph L Cooper, Morgan Quigley, Julie A Adams, [15] Orville Wright and others. 1977. How we made the first and Curtis Humphrey. 2008. Supporting wilderness flight. Department of Transportation, Federal Aviation search and rescue using a camera-equipped mini UAV. Administration. Journal of Field Robotics 25, 1-2 (2008), 89–110.