* Moved makeWhiteNoiseVector() to random.Normal.normal()
* Moved isControllable() and isDetectable() to system.LinearSystemUtil
* Renamed makeCostMatrix() to costMatrix() (Java)
* Renamed makeCovarianceMatrix() to covarianceMatrix() (Java)
* Renamed MakeCostMatrix() to CostMatrix() (C++)
* Renamed MakeCovMatrix() to CovarianceMatrix() (C++)
* Removed deprecated poseTo3dVector(), poseTo4dVector(), poseToVector()
* Removed clampInputMaxMagnitude()
* We don't use it, and Eigen has this functionality built in via `u =
u.array().min(u_max.array()).max(u_min.array());`
* Simplified implementation of desaturateInputVector()
The spiraling issue occurs when the vision rotation standard deviation is very high relative to the odometry rotation standard deviation and the vision measurements have a large rotation error. (Scaling the rotation component of a twist without scaling the translation component causes the direction of overall translation to change, leading to spiraling around (either towards or away) the vision measurement instead of moving towards it.) Using a transform instead of a twist avoids this issue.
In general, scaling twist components is more mathematically correct than scaling transform components. However, although twists are correct for modeling uncertainty in an odometry-only pose estimate, they are not correct for the difference between the odometry-only pose estimate and a vision measurement. Since neither twists nor transforms are completely correct (and the pose estimator as a whole is not mathematically correct), but using transforms can guarantee that the pose estimate approaches the vision measurement (instead of potentially spiraling away), they are the least bad option.
I upgraded all plugins I could see except org.ysb33r.doxygen. 2.0 made
breaking changes, and I couldn't figure out how to migrate.
Most of the changes are for suppressing new linter purification rites.
Adds methods to compute the dot and cross products between Translation2ds and Translation3ds, as well as methods to compute the square of Distance and Norm, which allows avoiding some calls to sqrt in many cases.
Co-authored-by: Tyler Veness <calcmogul@gmail.com>
Adds S3SigmaPoints based on MerweScaledSigmaPoints. In addition, restructures UnscentedKalmanFilter to support different sigma point generators and provides MerweUKF and S3UKF for convenience when working with either kind of filter.
S3UKFTest is copied from MerweUKFTest (which is a rename of UnscentedKalmanFilterTest). Curiously, however, in Java the original tolerance used in MerweUKFTest.testDriveConvergence() for the final rotation was too low for S3UKFTest, so the tolerance is increased from 0.000005 (5e-6) radians to 0.00015 (1.5e-4) radians. However, the C++ version still uses the original tolerance. (This difference is probably because Java uses a final rotation of 5.846 degrees while C++ uses a final rotation of 5.846 radians)
Closes#8072.
Breaking changes:
- (C++) UnscentedKalmanFilter has a new template parameter for the sigma point generator type.
- (Java) UnscentedKalmanFilter has an additional parameter to every constructor providing an instance of a sigma point generator.
- (C++) int MerweScaledSigmaPoints.NumSigmas() has been replaced with constexpr int MerweScaledSigmaPoints::NumSigmas.
- (C++) The second parameter of SquareRootUnscentedTransform has been changed from States to NumSigmas.
Ensures java deprecated notation is paired with javadoc and vice versa.
Adds javadoc deprecation for MecanumControllerCommand, ArmFeedForward,
ElevatorFeedforward, and MecanumDriveMotorVoltages
Fixes#7736
Supersedes #7737
Co-authored-by: Joseph Eng <91924258+KangarooKoala@users.noreply.github.com>
SimpleMotorFeedforward::calculate(velocity) was not updated to account for the removal of calculate(velocity, acceleration), so it would pass currentVelocity = velocity and nextVelocity = 0, resulting in negative outputs in many scenarios.
Reduces nesting by returning when the value is within the deadband.
Adjusts the algorithm to handle large values of maxMagnitude naturally (instead of needing a separate check).
Reformats the math comments.
Throughout the code the state sqrt covariance S and innovation covariance Sy are maintained as upper triangular cholesky factors of those covariance matrices. The original paper defines P=S*S', so S should be lower triangular. The functions in the paper reflect this. In the code implementation, the sqrt covariance matrices are upper triangular, but the algorithm expects them to be lower triangular.
This bug was likely missed because the incorrect version of the filter is able to converge for some systems where all the states are observed, and the test case is set up such that all states are observed.
To fix the bug, a couple things needed to be changed:
all instances of rankUpdate() needed to be changed to use the lower triangular cholesky factor,
In the unscented transform, when S is found via QR decomposition, we need to take the transpose because R is upper triangular,
P() and SetP() functions need to be modified to be P=S*S' instead of P=S'*S, and P.llt().matrixL() instead of P.llt().matrixU() respectively.
Each part of the algorithm has also had the comments changed to clarify exactly which equation from the paper it implements.
Co-authored-by: Tyler Veness <calcmogul@gmail.com>
