- reexport KiteUtils
- bump
KiteUtils
to v0.9.1 - when executing
bin/run_julia
, always executeusing KiteControllers
before displaying the REPL
- fixed logging of the height and X, Y and Z
- plotting of the height is fixed in
autopilot.jl
- the menu with the examples can now started by typing
menu()
- use the latest version of KiteModels, which defines the azimuth angle and the orientation differently and make the controllers and examples work with the new definitions
- bump
KiteUtils.jl
to v0.9.0. The new version has new fields that are used for logging. - the constructor
SystemStateControl()
now needs the additional parameterv_wind
- the constructors
WCSettings()
,FPCSettings()
andFPPSettings()
now have the new argumentupdate
. If true, then the settings are loaded from the correspondingyaml
file. - do not use the function
update_sys_state!()
any longer because it is buggy
- add the script
parking_wind_dir.jl
that tests the parking controller when the wind direction is changing - add the script
parking_controller.jl
which implements a dual-loop parking controller. The inner loop controls the turn rate. It has an excellent performance. - add the script
test/menu.jl
which allows to execute the manual tests, that display plots and fix the tests
- the script
create_sys_image
is now installing matplotlib if required - removed calls to se() to be sure the correct settings from the variable set are used
- fixed the wrong polars in the yaml settings for the 20 m² kite
- use the new, correct methods for calculating the polars in autopilot.jl
- added the new fields needed for the new winch controller release
- bump KiteUtils to 0.7.9
- bump KiteModels to 0.6.6
- fix all failing test scripts
- fix most of the example, and the script
menu.jl
to run the examples interactively
- bump KiteUtils to 0.7.4
- bump KiteModels to 0.6.3
- fix some examples and the script
create_sys_image
- explain different installation methods in README.md
- bump KiteUtils to 0.7.2
- bump WinchModels to 0.3.2
- bump KiteViewers to 0.4.16
- adapt WinchController to new WinchModel interface
- fix tests
- fix control_plot on Windows (added sleep)
- fix cycle counting in FPP
- add function plot_timing2()
- add lower limit for C2 of 2.0
- bump KiteViewers.jl to 0.4.14
- upgrade GLFW to latest version (the original issue is fixed now)
- downgrade GLFW_jll to fix an issue on Ubuntu 24.04 and on Mac
- add section kps4-3l to all settings files to be compatible with the latest KiteUtils package
- copyright disclaimer from TU Delft
- fix #35, save the default log file in the output folder
- add the field
corr_vec
to allfpp_settings_xxx.yaml
files - modify
kiteobserver.jl
to use this correction vector - the projects
hydra10_951
andhydra20_600
hydra20_920
work correctly now; key change: set k_c2_high to 6.0 - new plot control_plot_II
- remove compatibility with Julia 1.9, version 1.10 is the only supported version now
- use 60 Hz refresh rate for 3x, 6x, 9x and 12x time lapse
- refactoring: add type KiteApp to the script autopilot.jl
- make use of log_level settings, print a lot less for log_level=0
- refactoring: split
flightpathplanner.jl
inflightpathplanner.jl
andflightpathcalculator.jl
- created the simplified components
flightpathplanner2.jl
andflightpathcalculator2.jl
;
they are easier to understand and contain a lot less hard-coded constants
- adapt time_lapse according to menu selection
- adapt text_mod such that the text update rate stays constant
- make it possible to change
t_sim
using the GUI - add the plots
plot_elev_az2
andplot_side_view2
that omit the first power cycle - add the plots
plot_elev_az3
,plot_side_view3
andplot_front_view3
that omit the first two power cycles - add the plot
aerodynamics
that plots the lift-over-drag of the kite - add time series of
fig8
toplot_control
- add the file
kiteobservers.jl
which provides the typeKiteObserver
and methods to determine deviations from the desired flight path - add the script
learning.jl
that provides the methodtrain()
that implements iterative learning of a vector of flight path corrections
- the new, pure Julia solver DFBDF is now the default. It is on average 4 times faster, uses only half the memory and is much more stable. In my tests, for rel_tol
$=0.0005$ the solver induced error of the peak tether force and the harvested energy was always$<0.1$ %.
- the example
autopilot.jl
was vastly improved - a menu with 6 pre-defined plots was added to the GUI
- a statistics dialog was added to the menu
- saving and loading of log files added to the menu
- a second menu was added that allows to change the tolerance of the solver