Non-linear Filtering Techniques for Improving Accurate Vehicle Angle Determination

authored by: Richard Pfleiderer, Jannes Ramsdorf, Yannis Boländer, Holger Blume
Abstract: Accurate measurement of vehicle angles, particularly pitch angles, is crucial for the safety and efficiency of modern vehicles. These angles affect vehicle dynamics such as acceleration and braking and are essential for adaptive headlight systems that adjust lighting based on vehicle pitch. Various methods exist for measuring vehicle pitch angles, including LiDAR, wheel sensors combined with inclinometers, and inertial measurement units or accelerometers in conjunction with wheel sensors. A cost-effective approach employs an algorithm that calculates real acceleration from vehicle speed and accounts for road inclination using only accelerometers. However, this algorithm only provides reliable results with accurate accelerometer data. Since accelerometers are prone to noise, signal processing requires filtering. Linear filters, which significantly smooth the signal, are unsuitable here as they can distort crucial signal information through over-smoothing. Instead, nonlinear filters offer a better solution. The Kuwahara filter, total variation filter, and bilateral filter are such nonlinear alternatives that reduce noise while preserving essential signal details, such as edges and features. Tests indicate that without adequate filtering, the average deviation in pitch angle is 1.4 degrees, with a maximum error of 19.4 degrees. With the use of the bilateral filter, this average error is reduced to 0.11 degrees and the maximum error to 1.2 degrees.
Organisation(s): Institute of Microelectronic Systems
Architectures and Systems Section
Type: Conference contribution
Publication date: 2024
Publication status: Published
Peer reviewed: Yes
ASJC Scopus subject areas: Computer Networks and Communications, Computer Science Applications, Geophysics, Electrical and Electronic Engineering, Instrumentation, Radiation
Electronic version(s): https://doi.org/10.23919/ieeeconf64570.2024.10739191 (Access: Closed)