This reverts the addition of a check against mission items that have the
same position. This breaks existing MAVSDK implementations where a
LOITER_TIME item is set right after a WAYPOINT with the same
coordinates. It is an interim hack to allow the vehicle to hold still
during a photo is captured.
This leaves the check in place for gates where we need to be able to
calculate the direction between gate and waypoint.
- create common uORB::PublicationBase
- uORB::PublicationQueued types are now type aliases
- ORB_PRIO use enum type everywhere to avoid accidental misuse
- PX4Accelerometer/PX4Gyroscope/etc driver libs explicitly advertise on construction, unadvertise on destruction. This is a workaround for any potential issues that might appear when accel/gyro cdev and uORB indexing doesn't align.
This is a corner case that happens when the user wants to command an
altitude change just after a takeoff. Takeoff doesn't set the current
lat and lon therefore the user had to change altitude twice to get it working.
The first time to set the current lat and lon and the second time to
actually change altitude.
Before this check we were free to fly outside of a geofence with a goto
command. With this check we ignore the command if the location of the goto
is outside the fence. If this is the case we send back a mavlink warning to
the ground station.
This is an attempt to fix an edge case in the triplet publication which
can lead to crashes on autopilots with slow SD cards.
The sequence of events before this patch is:
1. Switch to POSCTL when disarmed. At this point current valid with
setpoint idle is published.
2. Arm, takeoff, and fly using joystick/RC.
3. Switch to RTL (or trigger RTL using RC loss). At this point the
setpoint is valid but still idle and the motors will shut off.
4. Once navigator has published the new setpoint (which can take up to
1.5 seconds on slow SD cards) we will hopefully recover.
With this patch we omit this edge case, so we never publish this idle
setpoint when landed. The assumption is that this idle setpoint is no
longer required with the current flight task code, however, that needs
to be further verified.
The mission logic depends in a number of locations on being able to calculate the direction from one waypoint to another. Missions that have waypoints that are in the same physical location do not make sense and need to be rejected (the GCS / SDK generating them needs to be fixed). By enforcing this we can work with a reasonable and simpler state machine while executing the mission.
A mission with zero elements is invalid but should not lead to a warning. Previously a fixed wing aircraft without a mission did emit warnings about a missing land item.
This enables the navigator to wait for a specific gate coordinate to pass orthogonally to the current trajectory. This is particularly helpful for payload / camera handling in missions and avoids a dependency of payload handling on navigation waypoints.
Prior this fix VTOL missions get rejected if they have a DO_LAND_START marker, but then do
not end with a LAND waypoint but a VTOL transition and land.
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
This avoids premature state machine execution, most notably for missions.
In particular this fixes these issues:
- when landed, and the vehicle has a valid mission: switching to mission
would execute the first item.
- after executing a mission with a land waypoint, the mission is reset
and the first item is executed.
In case the first waypoint is a takeoff, it would print a user message:
"Takeoff to x meters above home.". If it was a vehicle command, it would
execute that.
Note that because the vehicle is disarmed, the vehicle would not actually
do anything (except for commands that are accepted while disarmed).
* Treat UAVS diffrently from manned aviation
* Added fake_traffic testing functionality,
* Added NAV_TRAFF_AVOID Hold and Landmode
* Added Behavior: HOLD Position to collision avoidance mode and implemented Landmode to collision avoidance.
Boards where no Hardware GUID is defined will send 0 as GUID.
Right now collision avoidance for more than one FMU without Hardware GUID is not possible.
We should consider adding a randomly generated HW GUID as a placeholder for legacy Boards
Hamish Willee