In Copter with minimize feature set the entire AP_PathPlanning feature including the database are unavailable
Rover has enough space for the database even with minimise features set
AP_OABendyRuler: support exclusion polygons
AP_OADijkstra: support exclusion polygons
AC_Avoid: adjust_velocity supports exclusion polygons
AC_Avoidance: handle fence::get_boundary_points returning nullptr instead of setting num_points to zero
AC_Avoidance: Dijkstra's works with only exclusion polygons
AC_Avoidance: Dijkstra: check for fence counts instead of polyfence validity
We really only care whether fences can be returned - and they won't be
returned unless they are valid
AC_Avoidance: BendyRuler: just try to get inclusion fence rather than checking validity
AC_Avoidance: BendyRuler supports exclusion circles
AC_Avoid: Dijkstra support for exclusion circles
AC_Avoid: BendyRuler support for inclusion circles
AC_Avoid: stop an inclusion/exclusion circular fences
AC_Avoid: stop at inclusion/exclusion circular fences
AC_Avoid: fixes to Dijkstra's use of inclusion/exclusion circles and polygons
AP_Avoidance: take semaphores when interacting with AHRS and polyfence
AC_Avoid: Dijkstra's fix for some_fences_enabled inclusion circles
implements both bendy ruler and dijkstra's algorithms
includes many changes after peer review including:
removed unused airspeed and turn-rate arguments
renamed constants to start with OA_BENDYRULER_
removed fence_ptr check from update function that could cause bendy ruler to fail when fences were disabled
AC_Fence's fence-margin is not used and instead we always use a hard-coded 2m limit
vehicle's heading is used instead of ground-course when speed is below 0.2m/s
static assert added to ensure OA_BENDYRULER_BERAING_INC constant is defined correctly
replace BAD_MARGIN/GOOD_MARGIN with FLT_MAX/-FLT_MAX
use calc_margin_from_xxx function's return values
calc-margin-from-proximity-sensors fails if copy-locations fails
remove debug
remove unused singletons
and other changes including:
integrate fence rename to get_boundary_points
adapt to fence return point and duplicate final point from not being returned
create bendyrule and dijkstra objects on init
add pre-arm-check to ensure they've been created successfully
Dijkstra's detects polygon fence changes
minor fix to Dijkstra's return path
allow update_visgraph to accept 0,0 as extra point
and these fixes from peer reviews:
remove unused methods
add note about dangers of operator[]
reduce memory consuption of ShortPathNode
remove unused methods including print_all_nodes and print_shortest_path
replace OA_DIJKSTRA_POLYGON_VISGRAPH_INFINITY_CM with FLT_MAX
update method gets check that ekf origin has been set
fix indexing bug in inner fence creation
fix whitespace error
find-node-from-id uses switch statement instead of 3 IF statements
update_visgraph comments added to clarify usage of extra_position argument
find-closest-node-idx always uses node_index instead of int16_t
static assert that OA_DIJKSTRA_POLYGON_FENCE_PTS < 255
some looping style changes
AP_OABendyRuler: fix to ground course calculation
AP_OAPathPlanner: pre_arm_check writes to caller's buffer
remove unnecessary hal and stdio.h includes
rename fence_ptr local variable to fence
add sanity check to OADijkstra's set_fence_margin
add sanity check to OABendyRuler's set_config
BendyRuler's test_bearing local constants changed to be float constants
BendyRuler's ahrs_home made local constant
BendyRuler's proximity margin calcs lose margin_found local variable (use FLT_MAX instead)
BendyRuler's calc-margin-from-circular-fence performance improvement by removing one sqrt
AP_OABendyRuler: fix probe directions to be 5deg offsets from dir to destination
AP_OADijkstra: fixes after peer review
remove unnecesary initialisations
replace safe_sqrt(sq()) with length()
remove unnecessary block
comment fixes
AP_OAPathPlanner: add logging of final an intermediate location
the simpler stop feature (as opposed to slide) was not including the margin when looking for intersections with the boundary meaning it could slow too late
See discussion here:
https://github.com/ArduPilot/ardupilot/issues/7331
we were getting some uninitialised variables. While it only showed up in
AP_SbusOut, it means we can't be sure it won't happen on other objects,
so safest to remove the approach
Thanks to assistance from Lucas, Peter and Francisco
Rishabh
Patrick José Pereira
Leonard Hall
Iampete1
Peter Barker
Peter Hall
Andrew Tridgell
Randy Mackay
Tom Pittenger
JIMWAR
murata
Dr.-Ing. Amilcar Do Carmo Lucas
khancyr
Lucas De Marchi
Shingo Matsuura
Jacob Walser