
Work Plan
i4Driving is being conducted in the form of eight work packages (WPs):
WP1: Gathering evidence on Human Factors from data (led by TUM)
The objective is threefold: (a) to gather and harmonize existing data for model identification, calibration, and validation in WP2 and WP4; (b) formulate and structure hypotheses as a basis for the experiments in WP3 (driving simulators) and WP5 (field lab); (c) select a comprehensive set of use cases and scenarios that will be used throughout the project.
WP2: Robust 4D human driver models under uncertainty (led by TU Delft)
The objective is to develop a software library of validated mathematical models to simulate tactical and operational driving behaviors. It includes different types of lane-changing (LC), car-following (CF), gap assessment and acceptance (GA) models. Models will be used in classic traffic simulation, and in 3D virtual worlds (driving simulator experiments, metaverses).
WP3: Mapping human factors into driving performances through driving simulation (led by VTI)
The objective is twofold: (a) to study hypothesis and investigate variance of human driving behaviour in different situations, while respecting gender, age, driving experience, and other factors. That is to study the heterogeneity of human driver behavior in different near-critical scenarios (i.e., combinations of different types of drivers, use cases and road environmental conditions); (b) to investigate to what degree a model of human driving behavior is perceived as an actual human driver, in a driving simulation environment.
WP4: Encoding the heterogeneity of human and external factors into probabilistic driver behavioural models (led by UniNA)
The objective is to encode the heterogeneity of both human and external factors (e.g., gender, ageing, driving experience or weather conditions) resulting from naturalistic driving studies (NDS) and driving simulation experiments (DSE), into probabilistic distributions and correlation structures of model parameters, for the developed library of 4D human driver behavioural models.
WP5: Field testing of human behaviours in urban and freeway ODDs (led by TUM)
The objective is to collect data on human driving behavior in a near-to-real-life environment. The purpose of this work is twofold. In the first phase, to collect data necessary for informing the development of the model itself (especially on the impact of parameters that describe the driving task complexity). This means testing various hypotheses and assumptions before they are integrated into the model. Secondly, after model development, further tests are performed as part of the model calibration and validation process. To recreate certain near-collision scenarios safely, it may be necessary to incorporate augmented reality (AR) technology into the studies. This allows the test subject to interact with a mixture of real and simulated road users, expanding the range of possible testing scenarios.
WP6: Evaluating 4D human driver models in target applications (led by WMG)
The objective is twofold: a) to define a general evaluation framework, which develops and extends ADS evaluation frameworks to a wider range of applications, and b) to use the evaluation framework in five target applications. The framework combines scenario-based and traffic-based simulation approaches, and is composed of three elements: i) scenarios (incl. generation, description, format and storage), ii) environment (incl. identification of simulators, and scenario execution), iii) certification/analysis (incl. the assessment on whether the intended scenarios occurred and assess the scenario against criteria). Demonstrators of the probabilistic i4Driving library of models are set up, covering the following five targeted applications: Definition of the human driver baseline 1) for road safety assessment in PEARS; 2) in consumer testing campaigns of ADS; 3) for insurance qualification scheme design. Mixed traffic simulation (AVs and human-driven vehicles) for 4) ADS testing, and 5) Safety/traffic/energy efficiency evaluation.
WP7: Reproducibility, dissemination and exploitation (led by WMG)
The objective is to increase the impact of i4Driving by increasing the understanding and awareness of project aims and outcomes amongst a diverse set of stakeholders from the research community to industry and policy makers. One of the key focus areas is to ensure exploitation of the human driver models by regulatory bodies, consumer testing organisations and the insurance industry. It also ensures that all results (in some form) have OpenAcess and are available for the wider CCAM ecosystem. Along with OpenAccess publications, developed code, toolchain and models are open source.
WP8: Project coordination and management (led by PANTEIA)
The objective is to assure an effective coordination among all Partners, and to carry out a smooth and accurate financial, administrative and technical project management. Furthermore, it manages the (internal) relation and communication between the team and the European Commission.