Create Your own Drone.

"CREATE YOUR OWN DRONE "

   Build Your Own Drone Air  &  Space  Magazine   Ed Darack The author flies his homebuilt UAV, the Kestrel-6. (Ed Darack) In  2005,  I  was  embedded  as  an  independent journalist with  a  platoon  of  Marines  with  an attachment  of  Afghan  security  forces  in the  Hindu Kush,  just  a  few  miles  from  the  Pakistani border.  I  was  about to  step  into  a  possible  death  trap,  and  I’d  never  felt more  scared  in  my life.  “There  are  50  Taliban  massed  behind  that  ridge,”  said  “Sultan,”  our  Afghan  interpreter (his  real  name  is  being  withheld  for  security  reasons).  He  had  picked  up  chatter  from  his  handheld  radio.

“They  know  we’re  here,”  he  continued,  “and  that  we  have  to  cross  that  field.” “Time to get moving,” a nearby Marine said. We  began  to  move,  navigating  the  terraces  on  the  outskirts  of  a  small  village.  An  hour later,  we  regrouped  by  a  mosque.  No  attack  came;  we  later  learned  that  the  “chatter”  was  one person  transmitting  fake  orders  to fake  fighters—a  common  ploy  to  confuse  U.S.  and  coalition troops.  Back  at  the  Marines’  base,  I  asked  about UAVs  (unmanned  aerial  vehicles)  the  Marines could  have  used  for  reconnaissance  in  that  situation  to  confirm  the  warning,  had  fighters actually been gathering to attack. “Yes,  we  have  them,”  said  First  Lieutenant  Patrick  Kinser,  the  platoon  commander.  “Never use  them,  though.  They  often  end  up  being  recovery  operations—sometimes  dangerous recovery  ops.”  Kinser  was  talking  about the  two  Dragon  Eye  UAVs  that his  platoon  had  been given—and  had  to  chase  down  a  number  of  times  in  the  steep  mountains  around  the  base, after  they  crashed.  “And  the  image  quality  is  terrible.”  Kinser  said  that  even  the  slightest mountain  winds  made  the  UAVs’  video  feed  too  jerky  to  identify  much  of  anything  on  the ground.  “Can’t  make  out  anything  of  use,  really,”  he  continued.  “I  once  joked  we should  just fly them both into a tree, so we’d never again have to chase them down.” The  Dragon  Eye,  while  slightly  larger than  a  truly  “backpackable”  UAV,  was  also cumbersome  to launch:  Either it  required  a  bungee  cord,  or  one  of  the  grunts  had  to  throw  it hard  while  running,  exposing  himself  to  potential  enemy  fire. *** Five  years  after  my  experience  in  Afghanistan,  I  visited  the  Marine  Corps  Mountain  Warfare Training  Center  in  California’s  Sierra  Nevada  mountains.  I  wanted  to  know  if  the  instructors had  any  experience  with  a  small  UAV  that  was  capable  of  enduring  mountain  and  desert  flight. They  had  not,  although  they  mentioned  the  Dragon  Eye,  and  the  related  RQ-11  Raven,  but noted  that  neither  was  compact  enough  to  be  backpackable.  They  had  hopes  for  micro  air vehicles that they’d read about, but most were fixed-wing. “Think  of  urban fighting,”  said  Sergeant  Tony  Powers,  one  of  the  instructors  and  a  Marine scout  sniper.  “That’s  when  a  squad  could  really  use  a  micro  UAV  to  get  eyes  on  an  enemy position.  But  a  fixed-wing  platform  flies  too  fast,  and  isn’t maneuverable  in  those  tight confines.”  According  to  Powers,  the  most  useful  platform  for  such  situations  is  one  that  can move  in  any  direction,  quickly  or  slowly,  then  go  into  a  hover  and  pass  video  feed  back  to  the user—perhaps a micro-helicopter UAV. So  began  my  UAV project.  Like  aircraft  homebuilders  whose  requirements  couldn’t  quite  be met by  the  kits  or  plans  on  the  market,  I  wanted  to  construct a  UAV  optimized  for  a  set of tasks  that  my  experience  with  the  Marines  suggested.  Here  are  the  steps  I  took  to  create  the Kestrel-6,  named  after  the  bird  of  prey,  known  for  its  hovering  ability.  (The  “6”  came  from  the atomic  number of  carbon  and  the  number of  motors  used  on  the  craft.)  The  entire  project  cost roughly $3,500. Step 1: Age-old aviation question: Speed or loiter capability? (Ed Darack) I  chose  a  multi-rotor  aircraft  over  a  fixed-wing  or  helicopter  because  I  wanted  to  be  able  to get a  static  view.  Multi-rotor aircraft  are  also  fast,  very  stable,  and  able  to  launch  vertically  in the tightest of confines—even inside a room and out a window. Multi-rotors  typically  use  anything  from  three  motor-propeller  assemblies  to  eight,  mounted at  the  end  of  arms  that  are  centrally  interconnected.

I  chose  the  “Y6”  configuration,  composed of  three  motor-mount  arms  with  two  co-axially  mounted  motor-propeller  assemblies  at  the  end of  each  arm.  With  two  motors  per  thrust  point  (one  facing  up  as  a  “tractor,”  and  one  down  as  a “pusher”),  the  Y6  has  redundancy.  Because  it  has  only  three  arms,  it  gives  a  mounted  camera a  wide  field  of  view.  I  needed  the  UAV  to be  as  small  and  stable  as  possible.  Smaller fixedwing  UAVs  often  fly  “squirrelly,”  but  that’s  not  true  of  a  well-designed  Y6,  with  its  multiple points of thrust stabilized by a high-performance flight control computer. Step  2:  Better  be  brainy Multi-rotors  fly  with  remarkable  speed  and  stability  because  they  have  a  number  of  points of  thrust,  not  just  one.

Each  works  against  and  with  other  thrust  points—and  with  and  against gravity—to move the craft along three axes, and, if needed, hold it steady in one position. A  multi-rotor  pitches,  rolls,  yaws,  and  hovers  by  varying  the  speed  of  its  motors  (each connected  to a  fixed-pitch  propeller)  individually,  which  varies  thrust (for  pitching  and  rolling) and  torque  (for  yawing).  This  type  of  aircraft,  however,  is  inherently  unstable  unless “balanced”  by  a  very  powerful  flight control  computer,  one  that can  analyze  aircraft attitude and  position,  then  provide  control  inputs  (as  motor  speed rate  changes) orders  of  magnitude faster  than  a  human’s  ability.  Think  of  trying  to  balance  a  baseball  atop  the  tip  of  a  pencil: You’re  not  really  “balancing”  it,  but constantly  moving  the  pencil  under  the  baseball  in  a  dance with  gravity  to  get  a  few  brief  moments  of  relative  stability.  But most people  don’t have  the eye-hand  coordination  for  such  a  feat.  Similarly,  until  recently,  sensors  and  computers  simply couldn’t work fast enough to use multiple thrust points to control a small aerial vehicle. Over  the  past  few  years,  the  electronics  industry  has  made  great  strides  in  the development  of  micro-electromechanical  systems  and  inertial  measurement  units.  These include  tiny,  solid-state,  multi-axis  gyroscopes  for  spatial  orientation  and  accelerometers  to measure  change  in  velocity  to  guide  multi-rotor  and  other types  of  aircraft.  Manufacturers  also produce  micro-electromechanical  magnetometers  for  navigation,  and  pressure  sensors (barometers) for altitude determination. Multi-rotors  fly  with  remarkable  speed  and  stability  because  they  have  a  number  of  points of  thrust,  not  just  one.  Each  works  against  and  with  other  thrust  points—and  with  and  against gravity—to move the craft along three axes, and, if needed, hold it steady in one position. A  multi-rotor  pitches,  rolls,  yaws,  and  hovers  by  varying  the  speed  of  its  motors  (each connected  to a  fixed-pitch  propeller)  individually,  which  varies  thrust (for  pitching  and  rolling) and  torque  (for  yawing).  This  type  of  aircraft,  however,  is  inherently  unstable  unless “balanced”  by  a  very  powerful  flight control  computer,  one  that can  analyze  aircraft attitude and  position,  then  provide  control  inputs  (as  motor  speed rate  changes) orders  of  magnitude faster  than  a  human’s  ability.  Think  of  trying  to  balance  a  baseball  atop  the  tip  of  a  pencil: You’re  not  really  “balancing”  it,  but constantly  moving  the  pencil  under  the  baseball  in  a  dance with  gravity  to  get  a  few  brief  moments  of  relative  stability.  But most people  don’t have  the eye-hand  coordination  for  such  a  feat.  Similarly,  until  recently,  sensors  and  computers  simply couldn’t work fast enough to use multiple thrust points to control a small aerial vehicle. Over  the  past  few  years,  the  electronics  industry  has  made  great  strides  in  the development  of  micro-electromechanical  systems  and  inertial  measurement  units.  These include  tiny,  solid-state,  multi-axis  gyroscopes  for  spatial  orientation  and  accelerometers  to measure  change  in  velocity  to  guide  multi-rotor  and  other types  of  aircraft.  Manufacturers  also produce  micro-electromechanical  magnetometers  for  navigation,  and  pressure  sensors (barometers) for altitude determination. ack) I  settled  on  a  company  at  the  forefront  of  the  technology,  Hoverfly  Technologies.  Their HoverflyPRO  control  module  uses  16  parallel  processors  in  its  flight  control  computer to analyze  thousands  of  inputs  per  second  from  the  onboard  three-axis  gyroscope,  three-axis accelerometer, and digital pressure sensor.

The  controller,  a  printed  circuit board  that  measures  just 2.75  by  2.75  inches  by  0.5  inch high,  takes  flight control  inputs  from  a  digital  receiver  (taking  commands  from a  usercontrolled transmitter on the ground) and tells the Kestrel to go, stop, and hover. The  board  commands  the  camera  to  pivot up  and  down,  and  side  to  side,  has  an  altitudehold  function,  and  overlays  vital  flight data  on  live  video  fed  to  a  ground  station—if  a  video transmission system is mounted to the craft. I  also  bought  the  HoverflyGPS  control  unit,  which,  when  mated  to  the  PRO  board,  adds three-dimensional position hold, automatic return-to-home, and waypoint navigation.  5: Designing,  engineering,  and  constructing  the  UAV  body—with a little help from friends (Ed Darack) I  could  have  simply  mounted  all  of  the  components  on  a  pre-built  hobbyist  multi-rotor  body, but  I  wanted  the  final  aircraft  to  be  as  light  and  small  as  possible,  very  strong,  built  specifically around  my  components,  and  have  no  parts  such  as  wires  or  electronic  speed  controllers  (ESCs) dangling in the open. With  the  dimensions  of  all  of  the  components,  I  used  illustration  software  to  sketch  out the smallest craft possible. For  construction  material,  I  chose  carbon  fiber (which  I  ordered  from  DragonPlate,  in Elbridge,  New  York).  It’s  extremely  light and  strong,  although  notoriously  difficult to  cut,  and it’s an  electrical  conductor,  so  I  could  not  have  any  exposed  wires. Carbon  fiber  is also excellent at dampening  vibrations,  important for  any  flight controller,  as  vibrations  can  affect the  performance  of  the  accelerometer  and  gyroscope,  not  to  mention  the  video.

Then  came  the hard  part:  I  had  to  engineer  each  individual  piece  of  the  UAV  body,  but  I  had  no  experience  in engineering  or  computer-aided  design.  Most  of  the  professional  CAD  programs  cost thousands of  dollars,  far  out  of  my  price  range.  I  bought  an  older  version  of  TurboCAD  I  found  on  Amazon for less than $30. Using  exact  dimensions  of  the  HoverflyPRO  controller,  ESCs,  motors,  and  so  on,  I  engineered 29 parts in one month. A caliper micrometer with one of the extruded carbon fiber parts. (Ed Darack) Virtually  constructing  the  body  of  the  craft  with  TurboCAD,  I  was  able  to  include  foldingforward  arms  for ease  of  transport,  perfectly  align  all  holes  for screws  (I  wanted  everything  to be  connected  mechanically,  with  no  glued  parts),  and  gain  a  sense  of  how  it  would  balance with all components mounted—vital for multi-rotors. With  the  parts  engineered,  I  needed  them  milled  from  the  pieces  of  carbon  fiber  stock.  This turned  out to  be  much  more  difficult than  I  imagined.  Few  machine  shops  work  with  carbon fiber,  as  it “eats”  cutting  bits.  I  finally  found  Jason  Sauer  of  Pinnacle  Machining  in  Fort  Collins, Colorado, who agreed to help me. There  is  no universal  file  format  for  CAD;  Sauer was  quickly  able  to redraw  the  TurboCAD files  for the  simple  parts  by  hand,  but for  the  more  complex  pieces,  I  had  to  figure  out how  to convert  the  files. e  completed  UAV,  including  GoPro  camera.  The  landing  skids  keep  the  craft  high enough  off  the  ground  for  a  safe  launch  and  landing,  and  are  out  of  the  camera’s field of view. They’re also easily removed for transport. (Ed Darack) Stymied,  I  put  an  ad  on  Craigslist,  and  within  30  minutes  heard  from  Tom  Hanson,  a machinist-turned-engineer who has  his  own  firm,  Hardware  Collaborative.  He  was  intrigued  by my  project (he  often  donates  his  time  to  educational  engineering  projects),  and  converted  the files in minutes. With  the  files  completed,  Sauer  cut  all  of  the  parts—to  a  .0001-inch  tolerance—on  his  threeaxis Haas computer numerical-control milling machine. I  then  bought  an  assortment  of  black  anodized  hex  cap  screws  from  C  D  Fasteners,  and very-hard-to-find black anodized aluminum locking nuts from Fastener Express. Thanks  to  Sauer’s  skill,  the  pieces  of  the  craft fit  together  perfectly.  Carefully  cutting, soldering,  and  shrink-tubing  (to  insulate  and  protect  the  soldered  wires  and  connectors),  I finished the construction of the Kestrel-6. Step 6: Moment of truth (Ed Darack) Test  flying  the  UAV  was  the  mostfrightening  part  of  the  project.  As  each  part  was  one-of-akind,  I  thought  a  design  error  or  crash  would  kill  the  vehicle—and  the  entire  project.  And  crash it did, but not catastrophically—just enough to break some propellers. Then  it crashed  again,  from  about 30  feet  up.  Remarkably,  it  sustained  only  minor  damage, thanks to the bolted carbon fiber construction and the lack of exposed components. I  was  never able  to determine  what  exactly  caused  the  crashes,  but  it  could  have  been  a power  brownout.  The  crashes  actually  proved  to  be  a  good  thing,  as  they  demonstrated  the Kestrel-6’s resilience. To  ensure  that  I  had  the  craft configured  properly,  I  took  it  to  Bill  Clary  of  Got  Aerial,  LLC, based  in  the  Denver,  Colorado  area.  Clary  is  an  aerial  videographer  and  photographer  who shoots  from  a  variety  of  unmanned  multi-rotors.  Clary  fine-tuned  the  Kestrel-6  and  gave  me some  vital  piloting  pointers,  such  as  always  having  a  detailed  flight plan.  (The  Kestrel-6  runs out of battery power in about 12 minutes.) From  that point,  I  was  able  to  fly  the  Kestrel-6  with continuous  real-time  video  downlink  from the  GoPro,  which  I  could  position  wherever  I  needed  (and  use  to  record  high-definition  video footage).

Although  not  done  in  a  military  setting,  these  tests—including  up-close  footage  of  a  spinning wind  turbine  hundreds  of  feet  high  in  20-mph  wind—proved  the  concept:  Anyone  can  create  a small, rugged, purpose-built UAV and put it to practical use. Thanks  to  the  GPS,  I  was  able  to  “park”  the  craft and  hold  it at  a  chosen  altitude,  and  even have  it  fly  back  to  its  launch  point,  all  while  seeing  what  it  was  seeing,  and  easily  positioning  it for  other views—everything  I  wish  I’d  had  during  that  terrifying  moment  just before  entering the potential kill zone in Afghanistan.