BASELABS Create 3.2

Release notes

BASELABS is working on permanent product improvements including new features or bug fixing. We inform our users about software updates and releases regularly. Please contact us to share your experiences with our products and ideas for new approaches.

About

Efficient data fusion development for ADAS and automated driving
BASELABS Create is designed for the fast development of complex data fusion algorithms. BASELABS Create can be used with field-tested, pre-implemented algorithms as well as for the development of fully custom algorithms.

More information about BASELABS Create

UPDATES

This is a bug fix release of BASELABS Create which contains the following improvements:

Fixed a bug that could cause the BayesFilter.Predict() method to return wrong results

Problem: Shared mutable state in the Predict() method of the Bayes Filter class.

Description: Within the Predict() method of Bayes Filter class the probabilities of the samples have been updated in place (shared between the actual and updated/time propagated samples). The implementation could lead to numerically wrong results as the already updated probabilities are recursively used within the same time update cycle of the Bayes filter.

Steps to reproduce: Compare the results of the prediction step of the Bayes filter with the hand calculated results as explained, e.g. in Thrun et al. Probabilistic Robotics (MIT Press, 2005), Chapter 2, Exercise 2/3 or use the provided example 2.4.2 with non-identical initial probabilities.

Fixed: A local copy of the probabilities' array is used to keep the intermediate values of the probabilities before all the samples are updated. The newly calculated probabilities are only copied back when the iteration over all the samples is over.

Fixed a bug that could cause a biased calculation of the covariance in the SampleSet.ToGaussian() method.

Problem: ToGaussian() method and explicit cast to Gaussian of the SampleSet class.

Description: The implementation was using the normalization with N when calculating the estimate of covariance, where N is the total number of samples. The provided value is the estimate of the second moment around the mean under the assumption that mean is perfectly known. In the case the mean has to be also estimated from the sample set, the method provides a slightly over-confident estimate of the covariance.

Steps to reproduce: Call the method for an equally-weighted sample set of moderate size (e.g. 10 samples of 1D Space) and compare the numerical results to the ones from calling the standard covariance estimate from Matlab/Octave (cov() function) parameters or to the ones from hand calculations.

Fixed: The ToGaussian() method or, correspondingly, the explicit cast to Gaussian, are now using the normalization with N-1. The method provides the best unbiased estimate of the covariance when the mean is also estimated from the same sample set. The implementation also supports the sample sets with non-equally weighted samples (see documentation for the details of implementation).

Fixed a bug that could cause the PersistenceModel.Evaluate() method to return negative likelihood values in retrodiction scenarios

Problem: Track persistence model when used for retrodiction (i.e. propagation backwards in time).

Description: Time prediction of the persistence model with delivers numerically wrong results for negative time difference. This could lead to runtime errors for large negative time span and/or process noise values as the calculated likelihood values can become negative.

Steps to reproduce: Use Evaluate() method of the persistence model with negative time span and under condition that abs(Time x ExistenceProcessNoise) > 1.

Fixed: The Evaluate() method of the persistence model still accepts both positive and negative values of the time difference, but internally an absolute value of the time difference for existence prediction is used. The implementation ensures that the model delivers identical results for both for forward and backward time propagation.

Fixed a bug that could cause licensing issues on Linux

Solved issue with possible licensing failure due to race conditions under Mono .NET framework.

Contact & Information

Top of page
This website uses cookies to ensure you get the best experience on our website. By visiting www.baselabs.de you agree to the use of cookies.