• Transport to/from Kingaroy worked well for everyone
• Accommodation was suitable, despite the 2.4km separation. The 3-bedroom house had plenty of room for workshops to be set up
• Spares and tools were appropriate. No spares or tools left behind that we required
• RFD900 modems worked extremely well with a decent SNR margin (around 40dB) during all sections of the flight
• Camera and Imaging worked as expected
• Bottle drop accuracy was extremely good with the wind calculations taken into account.
• Failover between the 5.8 GHz and 900MHz links worked as expected
• Waypoint tracking during the mission was very accurate, despite the strong winds
• The “butterfly” overflight for Joe location refinement, wind estimation and bottle drop confirmation was very valuable and resulted in a much closer drop distance to Joe
• Packup after the flight went smoothly

• RC failsafe did not work during scrutineering. Some cables were loose too
• Team interview – equalise the speaking parts a bit more
• The trailer worked very badly when carrying the Porter. Due to the rough ground it was bouncing around significantly. Could have led to damage.
• Could have reduced the convoy to 3 cars (GCS van, car with airframe, car with people and spares)
• Ethernet switch was killed when it was accidentally connect to 12V power rather than a 7.5V UBEC
• Checklist had multiple issues

• Not enough practice with the checklist
• Confusion over who was supposed to do which check item
• Lack of communication between GCS and Pilot as to status of checks
• Confusion of the lack of separation from checklist items and procedural items

• No contingency plans for an aborted takeoff or landing
• No contingency plans for dealing with technical failures (such as the compass)
• Hard ground at the GCS made setting up the GCS tent (with pegs) difficult
• No method to level the antenna tracker on un-level ground
• Tent prevented visibility of the runway from the GCS operator stations
• Antenna tracker required a some yaw but mainly pitch offsets to correctly track the UAV
• Takeoff had many issues

• Incorrect compass offsets
• Not accounting for the strong crosswind
• Lack of experience in launching in strong crosswinds
• Abort procedures was too binary (STICK_MIXING was disabled)
• Lack of communication between Pilot and GCS in relation to wind conditions

• Matt’s GCS station had the incorrect (older) version of MAVProxy (due to last minute update of cuav repo for NMEA support)
• 5.8GHz antenna had too small a beamwidth, made tracking difficult
• Servo cables in the wings came out during roadtrip
• Due to dropped radio packets, the “joe move” command to move the butterfly pattern to a new position did not move all the appropriate waypoints the first time
• Confused communications between GCS, OBC Liaison and OBC Judges in regards to permission to return to airport
• Landing issues:

• UAV came in too fast due to too short approach
• Flare point was too far down the runway
• Lack of experience in landing in windy conditions
• Too much focus on gaining extra points in the OBC, rather than on safety

• No data on airframe safe limits (ie. max windspeed before operations are aborted)

• Checklists and procedures

• Make them clearer (who does what)
• Separate the checklist and procedure items
• Practice them regularly

• Abort modes and Contingencies

• Write up clearer procedures for aborted takeoff, landing, etc
• Practice them regularly

• Automated takeoff and landing

• APM code needs to be changed to be compass-independent
• Need to test in a variety of conditions
• Determine the maximum envelope for a “good” takeoff to guide saftey pilot decisions.

• Tracking antenna

• Perhaps replace with a couple of sector antennas
• Need a method to level the antenna mast independent of ground/van level
• Hard mounting of PixHawk may improve pitch accuracy

• Create a safety management system for CanberraUAV

• Checklists
• Procedures
• Tracking airframe and engine hours
• Abort modes and contingencies

• Ground Control Van

• Potentially move the GCS into the van itself to reduce setup complexity
• Integrate UBEC’s (don’t leave them separate) into any equipment connection to the 12V power system.

• Better develop communications between Pilot and GCS
• Practice more “full” setup (van, full GCS, etc) regularly with all team members on a regular basis.
• Have a set of speaking points (regularly updated) that team members can use when talking about CanberraUAV, to ensure all team members have a similar level of knowledge about our operations.
• Source Code Management

• Continous build and integration of MAVProxy, cuav etc. with email notifications of each successful/failed build. i.e. when Tridge or someone else makes a change, we all know about it and can update our own setups.
• Use the canberrauav git repo for deployment