- uncomment coordinate scaling which caused unwanted behaivor
especially that radius never got to 1 when joystick diagonal
resulting in loss of power/speed in curves
- now scaling radius only which is the desired functionality
- optimize debug logging
- fix joystick calibration config (invertion was mistaken)
- remove second fan instance since both fans are controlled via one pin
now
- more config options so fans turn on less at short movements
=> less noise and less relay cycles
Update connection plan with new wiring and pin assignment
- add display
- add temp sensors
- single fan
- add rotary encoder
Add wiring plan with details about hardware
Changed config and code for motor driver:
Now its possible to configure whether a and b pins are inverted separately
e.g. a connected to mosfets (inverted) and b connected directly
- add more debug output in joystick getData function
- update joystick config with new min/max values while testing
=> fixed bugged movement (joystick unusable) after winter
- bump esp-idf version since successfully compiled with 4.4.4
- slightly decrease msFadeDecel by 100ms
- different approach in button.cpp for case last button press was long
e.g. reworked commands:
- 1x long press = restart
- 1x short press = center stick, freeze
- 1x short 1x long = run new auto commands
- 2x short = toggle idle
- change default minOnMs / minOffMs for evaulated switch class -> fix
button presses sometimes not recognized
the joystick object is now available in the button class/task
thus can be used to trigger commands with certain combinations of
joystick and button
- not used yet -
auto.cpp:
- forward instructions from commands to newly passed pointer
control.cpp:
- handle instructions returned at generation of commands in auto mode
- add method toggleMode
- remove dupliate code in toggleIdle method
button.cpp:
- optimize command for enabling leg support, add instruction
auto.cpp:
apply fadeAccel and fadeDecel from commands to motors
main.cpp:
double stack size of button task (crashed controller every time)
button.cpp:
Modify test command to actually eject armchair leg support
control.cpp:
apply generated motor commands to motors
rename some variables:
typo: fadeAccel
more specific: auto_instruction_t
- automaticArmchair:
- add method addCommands for adding an array of
commands to queue
- add keys to simpleCommand struct
- fadeDecel (not used yet)
- fadeAcel (not used yet)
- instructions (new enum for running other commands in control task
, not used yet)
- button.cpp
- add example command to 1x button press (comment out previous cmd
temporarily)
- control.cpp
- update changeMode function: dont do anything when current mode is
already target mode
- Create class automatedArmchair in new files auto.hpp and auto.cpp
This class currently makes it possible to queue up and run commands
(motorstate, duty for certaion duration),
- add new file to cmakelists
- create automatedArmchair instance 'armchair' in config.cpp
- add loglevel configuration to main.cpp
- add AUTO mode to control.cpp (no way to change to mode yet)
When switching mode from MASSAGE to other mode while joystick input
frozen (button). It was still frozen when switching back to MASSAGE mode
later. This is confusing and is seen as a bug
-> reset locked input variable to false on mode change
New command and feature also made some general changes to control and button necessary
joystick.cpp:
- add optional parameter to joystick_CommandsDriving function
bool altStickMapping (default false)
if true this currently swaps BOTTOM_LEFT with BOTTOM_RIGHT for experimental
different joystick mapping
button.cpp:
- fix/move variable declaration outside of switch
- add 12x case: send count to control task
control.cpp:
- create new section in handle loop for button events
- move x1 commands from JOYSTICK and MASSAGE case to new button
section
- remove unnecessary variable buttonEvent (only using buttonCount now)
- add functionality to 12x button event -> toggle alternative stick
mapping
readme.md:
- add new x12 button command
button.cpp:
- Add command: reboot when long button press
- Add command: switch to mode JOYSTICK with 3x press
Readme:
- rework table with switch functions
- add new commands to table
- add drink holder to planned features
joystick.cpp:
- fix bug: missing 'break;' for switch case in function joystick_CommandsShaking
Fix bug where armchair moved too long every once in a while (especially
at start).
Minor adjustments of parameters.
main.cpp:
- set individual task priorities for each task e.g. fan or buzzer task has very low priority (did not have any immediate effect though while testing)
- increase handle interval of motorctl
change from 20 to 10 (same frequency as generation of massage
commands)
-> this fixed the bug with unexpected movement (motorctl could not
process every command from massage mode)
control.cpp:
- decrease delay in massage mode for more detail/levels at joystick
radius (from 20 to 10ms)
- increase fading in massage mode (400ms to 500ms) for slightly less hard
shaking
joystick.cpp:
- reduce max shaking amount
- swap modes
- top left shake backward
- top right shake forward
- bottom left/right shake rotating
Modified function that generates commands in massage mode (joystick.cpp)
- define variable stickQuadrant with axis as hysteresis from actual joystick data/pos
- prepare switch case for implementation of 4 modes (each quadrant)
- add basic content to modes for testing (different motor directions)
TODO: remove old code with pulsing shake variable and implement actual modes
In massage mode it is required that the motors react quicker, Added the
following functionality to control.cpp:
- when changing to MASSAGE mode the downfading is disabled and upfading is
reduced
- when changing from MASSAGE mode downfading is enabled and upfading reset
to default value
- add 3 functions to controlledMotor class in motorctl.cpp/hpp
- setFade: set specific msFade duration for acceleration or
deceleration
- setFade: disable or set to default ramp for acceleration or
deceleration
- toggleFade: toggle between fading enabled or disables for
acceleration or deceleration
- button.cpp
- add controlledMotor objects to constructor
- add new command 8x press which now toggles fade-deceleration of both
motors
- config.cpp
- slightly decrease fading durations
- Fix bug where case duty=0 was not handled in fading if structure thus no
incrementing action happened anymore (stuck at duty = 0)
- Set fade duration (acceleration, deceleration) to realistic value in
config
- Change fading algorithm to handle Acceleration and Deceleration for
forward and reverse separately (4 different cases)
- rename variables to make fading direction more obvious
e.g. msFadeUp -> msFadeAccel or dutyIncrementDon -> dutyIncrementDecel
TODO: fading needs optimization
- Fix bug where downfade did not work when driving forward
- Add BRAKE command functionality to motorctl (untested, since not used anywhere atm)
- Cleanup: Optimize code, add comments
!!!TODO: there is a fault in the current concept/logic:
fading up/down generally works, however when accelerating
REVERSE fade-down is utilized instead of fade-up.
=> Fix that bug, fade down should only be used when decelerating.
For fading-down and delay-between-state-change to work correctly when switching between motorstates
(e.g. FWD -> REV) the handle function was changed.
using a duty value -100 for reverse 100% and +100 for forward 100%
allows are more mathematical approach
changes to motorctl.cpp:
- convert state and duty to duty -100 to 100 when receiving target
command at start
- convert duty -100-100 back to state and duty at the end for applying
motor commands
With this changes the motors are faded down (same as already existing fading up)
when the target duty suddenly decreases.
It can be configured or disabled with setting msFadeDown to 0 in config struct of
motorctl object.
- Add new config parameter msFadeDown
- Change msFade to msFadeDown
- Add fade down functionality to motorctl instead of just setting the
duty to lower value
!!!TODO: fix major BUG when motor direction changes suddenly (before fade got to
0) the motor immediately starts in other direction with previous duty
(no/less fading up happening)
While testing this feature briefly that approach did not appear to be a good idea.
- noticed that the driving with joystick was a little unintuitive/unexpected (maybe just not used to it?)
- BUG that the state does not reset to center when previous state was
X-AXIS? Resulting in the armchair to slightly shake randomly at
joystick center
-> this has to be fixed if this approach is tested again.
This reverts commit b2b12fe5de1e16ea2f25dfa4b0c3ac6845024c71.
modified function
joystickPos_t joystick_evaluatePosition(float x, float y, joystickPos_t* prevPos)
because prevPos has to be stored http.cpp/hpp and joystick.cpp/hpp had
to be updated as well
TODO: add function without hysteresis again?
WIP this has to be tested
Use reference to joystick object in control class instead of accessing
the global evaluatedJoystick object
control.hpp:
- add joystick reference to constructor
- add local joystick pointer variable
control.cpp:
- constructor: copy pointer to joystick object
- use methods of joystick reference instead of global object
- update log output of timeout check
config.cpp:
- add joystick object to control object construction
- joystick.hpp/cpp
- add function joystick_scaleCoordinatesLinear(joystickData_t * data, float pointX, float pointY)
that scales the coordinates with 2 different slopes before and
after certain point
- control.cpp
- scale coordinates linear with new function instead of exponential in JOYSTICK mode
- scale coordinates linear with new function instead of exponential in HTTP mode
- config.cpp
- adjust / decrease http joystick tolerances
Note: tested the armchair in http and joystick mode briefly and
optimized the scaling point slightly
- joystick.hpp/cpp
- add function joystick_scaleCoordinatesExp(joystickData_t * data, float exponent)
which updates joystick data with its scaled coordinates and
re-calculated radius
- control.cpp
- scale coordinates exponential (pow 2) in JOYSTICK mode
- scale coordinates exponential (pow 2) in HTTP mode
- config.cpp
- fixed swapped x/y zero tolerances for hardware joystick
Note: tested armchair with scaling exponents 1.5; 2; 3 and noticed that
2 works best
- add function joystick_generateCommandsShaking that generates motor
commands from joystick data
- pulses motors:
- intervals depend on joystick radius
- direction depends on joystick position (currently only on-x-axis and on-y-axis
supported)
Update react web app:
- add dark background color
- change joystick colors
- increase joystick size by 50px
- change heading
- remove angle and radius
- not displaying anymore
- not sending to api anymore
Update http.cpp:
- remove radius and angle from json parsing
- limit radius to 1
As already did for http joystick:
- Add different config options tolerance zero for X and Y axis for
normal/actual joystick.
This makes it possible to set Y tolerance to a lower value resulting in
a more responsive turning action, with still having a large range around
X axis for turning mode
Fix several bugs noticed while testing the preceding commits in dev
branch:
- Fix bug in function scaleCoordinate
- scaling was wrong resulting in negative/inverted values at start of axis
- Adjust timeout value from 30s to 5min
- Fix http joystick behaivor
- calculate angle, radius and evaluate position AFTER the coordinates
have been scaled in control.cpp (bug introduced when switching applying tolerance
on controller instead of in the app)
- Add independent toleranceZero for X and Y axis -> unnecessary to have
large tolerance for x axis... makes turning more sensitive
- Fix stack overflow in control task
- controller crashed repeatedly when logging output was enabled in
control task
- -> doubled stack size at task creation in main.cpp
currently works, position hast to be evaluated AFTER coordinate scaling
- create new struct control_config_t with several variables previously
hardcoded in control.cpp
- modified constructor: add config parameter
- add definition of config struct in config.cpp
typedef struct control_config_t {
controlMode_t defaultMode; //default mode after startup and toggling IDLE
//--- timeout ---
uint32_t timeoutMs; //time of inactivity after which the mode gets switched to IDLE
float timeoutTolerancePer; //percentage the duty can vary between timeout checks considered still inactive
//--- http mode ---
float http_toleranceZeroPer;//percentage around joystick axis the coordinate snaps to 0
float http_toleranceEndPer; //percentage before joystick end the coordinate snaps to 1/-1
uint32_t http_timeoutMs; //time no new data was received before the motors get turned off
Add feature that switches to mode IDLE when duty of both motors did not
change over certain time.
control.cpp/hpp:
- add private function that handles timeout
- add public function that resets timeout
- add slow loop with timeout handle inside control handle loop
- joystick.cpp/hpp:
- move method scaleCoordinate from joystick class to public function
- modify scaleCoordinate function to accept float values instead of
ADC pin, change tolerance parameters to percent instead of absolute
number
- change method getData to use the public function now
- control.cpp:
- use scaleCoordinate function in http mode
- calculate radius in http mode
- config.cpp
- adjust tolerance thresholds for joystick to percent
- App.js
- disable "snap to zero" feature -> just scale joystick output to
value of -1 to 1
since the web-app runs successfully on the esp32 webserver now, the api
can be accessed directly - this makes the web app independent of the
dynamic controller ip
- remove ip from api url -> uses current host
- add testing code for changing a variable with a button
- comment out some debug output
