Spelling suggestions: "subject:"pendrive"" "subject:"vattendriven""
1 |
Automatic generation of critical driving scenarios / Automatisk generering av kritiska scenarierAlam, Mohammad Saquib January 2020 (has links)
Despite the tremendous development in the autonomous vehicle industry, the tools for systematic testing are still lacking. Real-world testing is time-consuming and above all, dangerous. There is also a lack of a framework to automatically generate critical scenarios to test autonomous vehicles. This thesis develops a general framework for end- to- end testing of an autonomous vehicle in a simulated environment. The framework provides the capability to generate and execute a large number of traffic scenarios in a reliable manner. Two methods are proposed to compute the criticality of a traffic scenario. A so-called critical value is used to learn the probability distribution of the critical scenario iteratively. The obtained probability distribution can be used to sample critical scenarios for testing and for benchmarking a different autonomous vehicle. To describe the static and dynamic participants of urban traffic scenario executed by the simulator, OpenDrive and OpenScenario standards are used. / Trots den enorma utvecklingen inom den autonoma fordonsindustrin saknas fortfarande verktygen för systematisk testning. Verklig testning är tidskrävande och framför allt farlig. Det saknas också ett ramverk för att automatiskt generera kritiska scenarier för att testa autonoma fordon. Denna avhandling utvecklar en allmän ram för end-to-end- test av ett autonomt fordon i en simulerad miljö. Ramverket ger möjlighet att generera och utföra ett stort antal trafikscenarier på ett tillförlitligt sätt. Två metoder föreslås för att beräkna kritiken i ett trafikscenario. Ett så kallat kritiskt värde används för att lära sig sannolikhetsfördelningen för det kritiska scenariot iterativt. Den erhållna sannolikhetsfördelningen kan användas för att prova kritiska scenarier för testning och för benchmarking av ett annat autonomt fordon. För att beskriva de statiska och dynamiska deltagarna i stadstrafikscenariot som körs av simulatorn används OpenDrive och OpenScenario-standarder.
|
2 |
HORN - Hank and OpenDRIVE Road Networks : An editor for creating HANK scenarios while working with OpenDRIVEÖberg, Kim January 2012 (has links)
HORN is a solution to the problem of how to implement scenarios in a more efficient way than was previously possible allowing researchers who wish to create scenarios for HANK the ability to quicker implement larger scenarios than was previously possible. OpenDRIVE is an open standard for road networks that is believed to be the way forward and Horn is an attempt at unifying OpenDRIVE scenarios with HANK - the driving simulator currently in use at Link\"{o}pings Universitet, thus futureproofing all work done to implement scenarios. Before HORN HANK scenarios were laboriously constructed with a really bad program or by hand and HORN tries to make the process far less painful. This thesis describes how to work with the Road Network Editor program HORN ("Hank and OpenDRIVE Road Networks") that was developed for working with HANK's scenarios as well as my experience implementing it and some of the fascinating rules for how to draw some exotic two dimensional geometries I found out about as I worked on HORN.
|
Page generated in 0.0865 seconds