Humatects offers a software library for the development, utilization and machine learning of driver models based on dynamic Bayesian networks: the Driver Modelling Suite (DMS).
Advanced driver assistance systems require models for monitoring, understanding, assessing and anticipating human drivers’ behaviors. Despite its importance, support for driver modeling has been limited so far. General-purpose suites such as R, MATLAB, Simulink and SCADE can be used for aspects of driver modeling, but lack the overarching toolchain to support rapid development from raw data toward executable models and tangible results.
With DMS, the user can efficiently develop driver models by reusing ready-made parameters and model structures for a variety of use cases, including driver intention recognition, traffic prediction and autonomous control. Developed using the feedback of automotive companies concerning unique automotive requirements and use cases, DMS comes with the following advantages: a variety of predefined, psychologically motivated parameters and distribution types; ready-made machine-learning algorithms for parameter and structure learning; tools and applications for data preprocessing and annotation; tools and applications for runtime utilization, visualization, evaluation and diagnostics of models and parameters; step-by-step example workflows to enable easy access for beginners; powerful API to provide complete freedom for experts; support of incremental learning of driver models, even during runtime; and smooth integration into third-party middleware (such as RTMaps) and driving simulators (such as SiLAB and Scanner).