Create a virtual learning environment to test and validate the behaviour of autonomous vehicles

Climate change and the need for renewable energy are driving the development of electric and hybrid vehicles, however, concerns about road safety still remain. To address this issue and provide better safety and increased mobility there is a need for the development of autonomous vehicle technology and now the automotive industry is heading towards bringing fully autonomous vehicles on the public roads in the next few decades. The major concern with these technology-driven vehicles is testing of autonomous vehicles on public roads as no human intervention would be allowed while driving and this may involve some risk for the driver and the surrounding environment as any error or fault in the system may lead to damage of that environment, loss of manufacturing cost, time, energy and even severe accidents could lead to loss of life. In addition, these vehicles consist of more complex designs than traditional vehicles and thus comparatively would require billions of miles of testing. Considering the above factors, the industry has come up with the solutions to test these vehicles in a virtual environment first using the software in the loop approach. This concept is still in development and therefore this paper aims to develop a virtual learning environment where the performance of the control algorithms for an autonomous vehicle can be tested and validated under different driving scenarios. Rigorous research was first carried out to find out the available testing methods and software for performing simulations using different algorithms imposed on the software model for object and path detection. Based on this review a modelling design approach was chosen to perform simulations in MATLAB software. Different driving test scenarios such as a roundabout and a parking lot were created in the Automated Driving System Toolbox and simulation was run in Simulink to test the behaviour of vehicle model in terms of Automated Emergency Braking, Lateral Control, Cruise Control, and results were observed and analysed in Bird’s Eye Scope view and in 3-Dimensional Environment using Unreal Engine. The Sensor Fusion technique was used to obtain more precise and accurate results. Vehicle dynamics of the model were also tested in order to compare the stability of the vehicle on the basis of the Kinematic and Dynamic Model respectively. The functionality provided by the software was fully explored and relevant results were presented. This paper is focusing on building a flexible virtual testing environment that can be easily deployed by SME’s and start-up companies to develop and test autonomous driving algorithms using the software in the loop approach