High Precision Outdoor and Indoor Reference State Estimation for Testing Autonomous Vehicles

Abstract

A current trend in automotive research is autonomous driving. For the proper testing and validation of automated driving functions a reference vehicle state is required. Global Navigation Satellite Systems (GNSS) are useful in the automation of the vehicles because of their practicality and accuracy. However, there are situations where the satellite signal is absent or unusable. This research work presents a methodology that addresses those situations, thus largely reducing the dependency of Inertial Navigation Systems (INSs) on the SatNav. The proposed methodology includes (1) a standstill recognition based on machine learning, (2) a detailed mathematical description of the horizontation of inertial measurements, (3) sensor fusion by means of statistical filtering, (4) an outlier detection for correction data, (5) a drift detector, and (6) a novel LiDAR-based Positioning Method (LbPM) for indoor navigation. The robustness and accuracy of the methodology are validated with a state-of-the-art INS with Real-Time Kinematic (RTK) correction data. The results obtained show a great improvement in the accuracy of vehicle state estimation under adverse driving conditions, such as when the correction data is corrupted, when there are extended periods with no correction data and in the case of drifting. The proposed LbPM method achieves an accuracy closely resembling that of a system with RTK.

Generation of Correction Data for Autonomous Driving by Means of Machine Learning and On-Board Diagnostics

“…Similar as in [ 9 ], the authors of [ 10 ] demonstrate that ML-algorithms can run in real-time, even in resource-limited microcontrollers. Another research work that uses ML-methods for the estimation of the vehicle state is presented in [ 11 ]. There, the authors train with real data a Random Forest to detect when a vehicle is standing still.…”

“…Without such a classifier, the continuous integration of the inertial measurements leads to an apparent ever-increasing velocity over ground, which in turn distorts the rest of the estimated state vector. The methodology of [ 11 ] also deals with the fusion of the inertial signals with vehicle motion models in order to improve the accuracy of the vehicle state during Dead-reckoning. More methods that combine mathematical modeling with ML-methods can be found in [ 12 ].…”

“…With respect to (3), the present work deals with the generation of velocity correction data, instead of accelerations and angular velocities. The difference is relevant because, as shown in [ 11 ], by directly generating the velocity correction data, one is able to improve the accuracy of the navigation solution when compared to the one obtained from the integration of inertial measurements. Finally, with regards to (4), the present work analyses as well the possibility of implementing the proposed methods for online processing.…”

Scientific Developments and New Technological Trajectories in Sensor Research

“…[ 33 , 34 , 36 , 43 , 45 , 46 , 49 , 50 ] (cf., also [ 28 , 47 , 48 , 54 , 116 , 117 , 118 , 119 , 120 , 121 , 122 ]). Results suggest that sensors have, as parasite technologies (i.e., depending on other technologies; [ 17 ]), a wide spectrum of applications in medicine, environmental pollution, aircraft and automotive industries [ 123 , 124 , 125 , 126 , 127 ]. Moreover, the success of smart sensors is associated with the integration of the Internet of Things, through which it is possible to connect devices and exchange information among people, systems, objects and many other devices [ 128 ].…”