Virtual Environment for Training Autonomous Vehicles

Author(s):  
Jerome Leudet ◽  
Tommi Mikkonen ◽  
François Christophe ◽  
Tomi Männistö
Author(s):  
Marc Compere ◽  
Garrett Holden ◽  
Otto Legon ◽  
Roberto Martinez Cruz

Abstract Autonomous vehicle researchers need a common framework in which to test autonomous vehicles and algorithms along a realism spectrum from simulation-only to real vehicles and real people. The community needs an open-source, publicly available framework, with source code, in which to develop, simulate, execute, and post-process multi-vehicle tests. This paper presents a Mobility Virtual Environment (MoVE) for testing autonomous system algorithms, vehicles, and their interactions with real and simulated vehicles and pedestrians. The result is a network-centric framework designed to represent multiple real and multiple virtual vehicles interacting and possibly communicating with each other in a common coordinate frame with a common timestamp. This paper presents a literature review of comparable autonomous vehicle softwares, presents MoVE concepts and architecture, and presents three experimental tests with multiple virtual and real vehicles, with real pedestrians. The first scenario is a traffic wave simulation using a real lead vehicle and 3 real follower vehicles. The second scenario is a medical evacuation scenario with 2 real pedestrians and 1 real vehicles. Real pedestrians are represented using live-GPS-followers streaming GPS position from mobile phones over the cellular network. Time-history and spatial plots of real and virtual vehicles are presented with vehicle-to-vehicle distance calculations indicating where and when potential collisions were detected and avoided. The third scenario highlights the avoid() behavior successfully avoiding other virtual vehicles and 1 real pedestrian in a small outdoor area. The MoVE set of concepts and interfaces are implemented as open-source software available for use and customization within the autonomous vehicle community. MoVE is freely available under the GPLv3 open-source license at gitlab.com/comperem/move.


2006 ◽  
Vol 532-533 ◽  
pp. 1128-1131
Author(s):  
Yan Fei Liang ◽  
Han Wu He ◽  
De Tao Zheng ◽  
Xin Chen

This paper established the framework of the decision-making model system for autonomous vehicles. Based on virtual reality environment modeling technology, the virtual scene was obtained. The driving performance of autonomous vehicles in real environment was simulated with that of the virtual vehicle in virtual environment. It was studied the influence of driver’s aggressiveness on lane-changed performance through considering human factors, and several longitudinal driving modes were classified and discussed. Three-power B spline function was used in this paper to plan path by interpolating characteristics points. The driving framework and the driving models described in this paper serve to address the problem of building more realistic traffic at the microscopic level in driving simulators. The autonomous vehicles based on this system can be used as the vehicles in simulators and help to design traffic or help to verify the performance of vehicles.


Author(s):  
Mohsen Malayjerdi ◽  
Vladimir Kuts ◽  
Raivo Sell ◽  
Tauno Otto ◽  
Barış Cem Baykara

Abstract One of the primary verification criteria of the autonomous vehicle is safe interaction with other road users. Based on studies, real-road testing is not practical for safety validation due to its time and cost consuming. Therefore, simulating miles driven is the only feasible way to overcome this limitation. The primary goal of the related research project is to develop advanced techniques in the human-robot interaction (HRI) safety validation area by usage of immersive simulation technologies. Developing methods for the creation of the simulation environment will enable us to do experiments in a digital environment rather than real. The main aim of the paper is to develop an effective method of creating a virtual environment for performing simulations on industrial robots, mobile robots, and autonomous vehicles (AGV-s) from the safety perspective for humans. A mid-size drone was used for aerial imagery as the first step in creating a virtual environment. Then all the photos were processed in several steps to build the final 3D map. Next, this mapping method was used to create a high detail simulation environment for testing an autonomous shuttle. Developing efficient methods for mapping real environments and simulating their variables is crucial for the testing and development of control algorithms of autonomous vehicles.


Author(s):  
Roger Woodman ◽  
Ke Lu ◽  
Matthew D. Higgins ◽  
Simon Brewerton ◽  
Paul A. Jennings ◽  
...  

Electronics ◽  
2021 ◽  
Vol 10 (6) ◽  
pp. 694
Author(s):  
Mingyun Wen ◽  
Jisun Park ◽  
Yunsick Sung ◽  
Yong Woon Park ◽  
Kyungeun Cho

Recently, virtual environment-based techniques to train sensor-based autonomous driving models have been widely employed due to their efficiency. However, a simulated virtual environment is required to be highly similar to its real-world counterpart to ensure the applicability of such models to actual autonomous vehicles. Though advances in hardware and three-dimensional graphics engine technology have enabled the creation of realistic virtual driving environments, the myriad of scenarios occurring in the real world can only be simulated up to a limited extent. In this study, a scenario simulation and modeling framework that simulates the behavior of objects that may be encountered while driving is proposed to address this problem. This framework maximizes the number of scenarios, their types, and the driving experience in a virtual environment. Furthermore, a simulator was implemented and employed to evaluate the performance of the proposed framework.


Author(s):  
Vimal Rau Aparow ◽  
Apratim Choudary ◽  
Giridharan Kulandaivelu ◽  
Thomas Webster ◽  
Justin Dauwels ◽  
...  

Sign in / Sign up

Export Citation Format

Share Document