Falcon Voice -
Voice Interface Design Sprint

The Falcon Voice interface was created during a 12 hour design sprint with one teammate.  The goal of the sprint was to design specifically for voice interaction, with an end result of a working voice prototype.  We selected to design in the area of aerial filmmaking, with a focus on flight and camera controls for semi-professional or professional filmmakers who use drones to film.  The voice interactions revolve around automating commands or changing settings that are too unwieldy to accomplish while the user's hands are occupied controlling the aircraft.  The multi-path prototype was built entirely in Adobe XD, and additional Adobe CC tools were used in creating the look and feel of the prototype.  The XD prototype can be viewed online with an active Adobe account, keeping in mind that voice commands are only active while the space bar is depressed:  Falcon Voice Design Sprint Prototype.

12 Hours (during NOV 2018)

Prototyping, UX Research, UX Design

Adobe XD, Gestural Hand Sketching (initial concepts), Adobe Photoshop, Semi-structured Remote Interview
User Research and Design
     To ground our design decisions in evidence I first consulted with two experts in the film and photography industries.  The first, a professional filmmaker, provided invaluable input in terminology frequently used during filming and lent his expertise in camera settings.  The second, a professional photographer and semi-professional drone-based filmmaker provided user input for drone piloting commands.  During the informal interview with the experts, I discussed the required functions to operate an aerial drone, discovered user pain points that commonly occur in operation, and deliberated on the viability of a voice assisted interface.  The following were insights that inspired our concept design:

1. Flight and Camera Multi-tasking: Users must pilot their drone in airspace with enough precision to capture desired shots while also controlling all camera functions.  Basic flight operations (take-off, landing, movement, altitude adjustments, etc.) require user attention at the exact same time as camera functions (tilt, pan, zoom, etc.) that frame shots.  
Design Idea: automate basic flight control with voice commands to allow user to focus on camera work.

2. Costly Camera Adjustments:  In current drone-assisted filming, making fine adjustments to camera settings such as aperture or shutter speed requires the user to consume valuable flight time navigating multiple menus.
Design Idea: voice-based interface takes in phrases such as “open aperture” and other complex camera commands to change settings automatically.

3. Flight Paths are Manually Piloted or Require Complex Setup:  Drone piloting is a mixture of manual and complex automated scripts.  Users often forego the use of complex scripts and opt for manual operation once in air due to the time required to setup scripted shots in the drone menu system.  However, in manual mode, the user is required to attentively pilot the drone while simultaneously considering camera framing.
Design Idea: voice-activated automated flight paths for drone based on scripted phrases, such as “orbit target at X speed.”

     After our abbreviated user research, we developed a set of requirements for our system design must possess.  First and foremost, the system must not interfere with the safe operation of the drone, both in the domain of the drone’s mechanical safety features and in the user’s awareness during operation.  Additional requirements include the following:
1. Voice interactions must include keywords to prevent a user from accidentally triggering the system during normal speech.
2. System must provide a easy way for a user to cancel or undo commands (easy error correction).
3. All commands must have an equivalent, manual option in the drone’s standard visual interface, to allow for redundancy in the event there is a problem with the voice system.
4. System must be easy to learn and allow for operation by professional and novice pilots.
5. Voice commands must provide shortcuts to key functionalities used by filmmakers, such as positioning the camera for a shot and adjusting the camera's settings.
Prototyping in Adobe XD
     With user research and design requirements serving as a foundation, my partner began his sketch and design work in Adobe Illustrator, and I began the creation of our XD prototype.  We knew that we wanted to present a free-flowing experience for users, while at the same time we knew that the prototype must remain simple to fit the 12 hour time constraint of the design sprint.  Simple clicks move the user to three types of experience, and voice commands allow the user to explore inside each experience.  The voice interaction flow can be best seen by manipulating the prototype but a short overview is included here.  The flow consists of three parts: command > confirmation > execution. 
1. Command:  A user speaks the keyword "Falcon," followed by a command sentence.  A finalized version of the system would provide the user with the ability to add dynamic variable settings (distance, speed, etc.) as part of a command.  In the prototype itself, interaction has been limited to showcase the experience, and only a simplified version of the commands (no dynamic variables) has been included.  These simple commands use amateur filmmaking terminology to allow for the widest understanding among users.
2. Confirmation: Once a command is received, the Falcon system would provide an audio and visual confirmation of the command it received.  The limited capability of Adobe XD and how XD handles trigger events prevents audio feedback in the prototype.  If we choose to pursue this design in the future, both audio and visual confirmation would be included in the next prototype iteration.
3. Execution: After the system provides confirmation to the user, the drone executes the command.  In the case of movement commands, the user is now free to operate the camera with their full attention.  In the case of camera commands, the user is free to focus on operating the drone.

     Adobe XD was an adequate choice to explore voice interaction, mainly due to the integrated voice capability they added in their update several months ago.  However, the program still has some room to grow, specifically with adding support for multiple triggers or multiple effects keying off of one trigger.  There is also a shortcoming with the lack of video or even .gif support, forcing time-intensive workarounds that still result in low fidelity animation effects.  XD has come a long way in basic micro-interaction animation for buttons and such, but there are definitely cases where designers are wanting to use full video in their prototype.  Lastly, although XD is an imperfect tool, it does have quite a bit of power to quickly explore concepts and display them to users in a way that can rapidly result in user feedback, which is the real benefit of this type of prototyping.

You may also like

Back to Top