During my master’s thesis I firstly present the industrial standards for driver assistance systems and autonomous driving. After explaining definitions from within the field of machine learning, I display the algorithms mostly used in the field, with primary focus on image processing. I explicate the various metrics that are used during the evaluation of artificial neural networks, including qualitative and runtime performance measures. I expose some publicly available datasets that can be used for autonomous driving systems, and their specialities. To create a fast multitask system, I propose a meta-architecture which has a cascaded structure, with a single encoder – multiple decoder module embedded. After implementing this system using the available toolset, I execute a detailed performance analysis to rate the proposed solution by the explained metrics. Based on the analysis results, I suggest further optimisation possibilities.