Guerritore, Martina (2024) LiDAR Systems for Advanced Assisted Driving in Tramway Sector. [Tesi di dottorato]

Anteprima

Testo Guerritore_Martina_36.pdf Download (16MB) | Anteprima

Abstract

n my thesis, I focused on developing an assisted driving system for the tramway industry using Light Detection and Ranging (LiDAR). Research objectives include identifying, and localizing objects and then, estimating tram position in unknown environments. Regarding the first objective, the proposed methodology involves background subtraction, pose estimation, and 3D bounding box fitting for detailed state analysis, aiming to assess collision risks and alert drivers. An approach balancing speed and accuracy is proposed to overcome hardware limitations in processing high-performance LiDAR data. The solution is tested on real data sets collected on a city road with a 360° LiDAR. The effectiveness of this solution is compared with the most widely cited solutions. The second objective introduces two approaches for estimating tram position and orientation. For the tram position, a sensor fusion approach is suggested, enhancing accuracy and reliability and ensuring continuous localization despite sensor malfunctions. Unlike several alternatives, the considered one operates without dynamic or error models, offering a low computational burden. Performance analysis is conducted in a simulation environment replicating an architecture with four kinematic sensors. The adopted method maintains the error in position measurement, derived indirectly from an accelerometer with a 1 ppm offset on the full scale, within a few ppm of the full-scale position. Instead, estimating orientation involves integrating gyroscope measurements, but integrating over extended periods accumulates errors, often addressed using complementary filters to fuse data from accelerometers and gyroscopes, refining the orientation estimate. To avoid accelerometer-related errors, a method to fusion point clouds and gyroscope measurements is proposed. Real data is utilized to evaluate the effectiveness of all proposed strategies in a real case study.

Downloads

Actions (login required)

Modifica documento