If a Sendable like SendableChooser is destroyed and recreated, it leaves
a stale object in the Sendable registry. Using this object results in a
crash. This patch avoids using the stale object.
We should remove stale objects from the global registry upon object
destruction, but this fixes the crashing issue for now.
Closes#2818.
Co-authored-by: Tyler Veness <calcmogul@gmail.com>
Previously this sent just the raw analog value; the scaled value is likely what users expect.
Co-authored-by: Corey Applegate <coreya@centralmcgowan.com>
- Add link to code of conduct
- Update clang version used to 10 to match CI
- Update pull request format to match current practice
- Change Azure references to Actions
This address some problems with the LinearSystemLoop class that were discovered through testing.
The initial state estimate of the observer was set to the provided initial state rather than zero as previously, a non zero initial state passed into reset() would lead to a discrepancy between the current state estimate and the actual system state.
The CMake enable/disable flags as currently structured are a confusing mix of
WITH, WITHOUT, and USE with odd defaults. This changes the flags to consistently
use WITH and default the build options to everything enabled.
Currently, StepTiming() advances the time by the given delta, then runs
any Notifiers that expired within that timeframe until their expiration
times are in the future. This doesn't reflect how the Notifiers would
actually run on a real robot. For example, if a Notifier measures the
time between calls for state-space model advancement, it would measure
a large jump in time once, then zero for subsequent runs until the
Notifier was caught up to the current time.
With this change, the time is incremented by the full delta or until the
soonest Notifier, whichever has the smaller delta, then Notifiers set to
expire at that time are run. This is repeated until the time has been
advanced by the full delta. For the state-space model Notifier situation
mentioned before, it would measure multiple small time jumps instead of
one big one.
Currently, teams have to make a Notifier to run feedback controllers
more often than the TimedRobot loop period of 20ms (running TimedRobot
more often than this is not advised). This lets users add callbacks to
the main robot loop that run at a user-defined period. This allows
running feedback controllers more often, but does so synchronously with
TimedRobot so there aren't any thread safety issues.
To make the tests reliable, the synchronization in simulation Notifiers
had to be reworked. StepTiming() now waits for all Notifiers to reach
HAL_WaitForNotifierAlarm(), then steps the time, then lets any expired
Notifiers run.
While there, we made some variable names more descriptive and added more
comments.
There were three bugs:
1. The input range variables used in ProfiledPIDController::Calculate()
weren't being updated
2. The modulus error calculation was incorrect.
3. The setpoint wasn't being wrapped like the goal, so the invariant
that the error remains less than half the input range was violated.
(Thanks to @CptJJ for pointing this out and suggesting a fix.)
By storing the previous dt, it can be moved into Correct() where it is
actually used. This lets us take the continuous R as an argument in the
user-provided R overload.
This does not introduce any breaking changes for teams that used the TrajectoryGenerator API for
quintic splines with poses.
The GetQuinticControlVectorsFromWaypoints() method was removed because it is not possible for us (or
would require breaking changes to the shape of the splines) to generate only one quintic control vector
per Pose2d. Users who actually have control vectors directly (i.e. not from Pose2d objects, but a
dashboard such as PathWeaver) should have the number of control vectors correspond to the number of
"waypoints" and can call GetQuinticSplinesFromControlVectors() directly.
