Driving simulator exams are preferred — for understandable explanations: any scenario can be simulated at the touch of a button. They are impartial of time and weather situations and with no any protection possibility for the motor vehicle, persons or the setting. Additionally, an hour in the driving simulator is less expensive and necessitates fewer corporation than a true driving lesson on a take a look at keep track of. “In the area of remarkably automated driving, having said that, driving simulator scientific tests are often questioned due to the fact of the lack of realism. In addition, until lately there had been no standardized take a look at treatments that could have been utilized to look at elaborate tasks these kinds of as the mutual interaction in between human and program (handover treatments),” states Arno Eichberger, head of the investigate place “Automated Driving & Driver Assistance Systems” at the Institute of Automotive Engineering at Graz University of Technological know-how (TU Graz).
New regulation as initial spark
Not too long ago, due to the fact the very first worldwide regulation for Automated Lane Retaining Systems (ALKS) has been in power since the starting of 2021. This legislation resolves the street acceptance dilemma, as Eichberger points out: “Till now, regulatory authorities did not know how to take a look at and approve autonomous driving devices. The motor vehicle suppliers, in turn, did not know what requirements the devices had to meet in get to be authorised.” In the polices, the acceptance criteria for remarkably automated devices (autonomous driving degree three) up to a maximum pace of sixty km/h have now been specified for the very first time on the foundation of a traffic jam assistant. When the assistant is activated, obligation for handle is transferred to the device. The driver might consider their hands off the steering wheel, but has to right away consider in excess of once more in the event of a malfunction. The program will have to realize that the human being guiding the wheel is capable of doing this.
Based mostly on this regulation, Eichberger and his investigate partners from Fraunhofer Austria, AVL and JOANNEUM Research have made an effective process in excess of the final couple months by which the readiness to consider in excess of handle can be examined safely and securely, successfully, and to a significant degree realistically in a driving simulator and the results can be utilized for the certification of ALKS devices.
Identical device perception of the setting
Procedures had been needed to confirm the validity of the driving simulation making use of the take a look at generate. The foundation for this was a direct comparison — driving simulation and true driving (the AVL take a look at keep track of in Gratkorn, Styria, served as the take a look at area) had to match as intently as probable. Listed here, the device perception of the setting posed a challenge. Figuratively talking, device perception is the sensory organs of the motor vehicle. It has the undertaking of exactly recording the vehicle’s environment — from the landscape and environmental objects to other street end users — so that the driving support program can react appropriately to the circumstance. Eichberger: “If this is to operate the very same as in actuality, the environments in the simulation have to match the true setting to the correct centimetre.”
Transferring the driving routes to the driving simulator
his precision is realized making use of so-called “Extremely High Definition Maps” (UHDmaps®) from JOANNEUM Research (JR), just one of the world’s major investigate institutions in the area of electronic twins. “We use a mobile mapping program to measure the take a look at environments. Lastly, a seamless 3D map with an really significant degree of depth is developed from the measurement data. In addition to traffic infrastructure objects these kinds of as traffic signs, lane markings and guard rails, vegetation and buildings are also represented in this map,” states Patrick Luley, head of the investigate laboratory for remarkably automated driving at the Digital Institute. While similar precision can be realized with manual 3D modelling, JR’s automated UHD mapping course of action is a lot of periods less expensive and speedier.
The significant-resolution 3D setting is at last transferred to the driving simulator. This is in which the Fraunhofer Austria workforce appear in. Volker Settgast from the Visual Computing enterprise device: “We prepare the data in these kinds of a way that the 3D setting can be exhibited at significant pace.” Even reflective and clear surfaces or trees and bushes blown by the wind can be perceived normally. Depending on the take a look at scenario, further automobiles or even persons can then be extra to the virtual setting.
The validation is at last confirmed with the assist of comparative operates on the true route. “With our process, it is probable for automobile suppliers to effortlessly review and validate a specific sampling on the true keep track of and in the driving simulator. This suggests that the take a look at can in the long run be transferred from the true keep track of to the driving simulator,” states Eichberger. The TU Graz researcher and his workforce are now functioning on location up virtual acceptance exams in excess of the upcoming couple months.
Resources provided by Graz University of Technological know-how. Unique prepared by Christoph Pelzl. Notice: Written content might be edited for style and length.