Determining the Embedded Key Performance Indicator (KPI), based on a Fuzzy FxLMS Algorithm

In this work, we present the evaluation of the algorithmic performance of perception, achieved by a video sensor, within the framework of the Highly Automated Driving (HAD) project of the Obuda University. We process the quality of the real-time uncertainty propagation in the embedded environment. The perception software could be characterized by algorithmic key performance indicators (KPIs) which is the measurable metric of the video sequences uncertainty. Based on endurance runs, for the real-time performance determination of the algorithms, an adaptive filter structure could be approximated. The employed performance architecture is a fuzzy-filtered (F-Fx) version of the LMS (Least Mean Square) algorithm in the embedded autonomous driving software platform. The F-FxLMS model is designed for the prediction of the embedded real-time filter outputs. The key performance indicator is related to the adaptation number of the F-FXLMS filter.