The overarching objective is to deliver a new library of credible models of heterogeneous human driver behaviors which provides a human road safety baseline for CCAM virtual assessment.
A new library doesn’t mean new models. It means a combination of models, suitable and valid for both scenario-based and traffic-based safety assessment, which bring the heterogeneity and complexity of the road traffic system into simulation. Adding sufficient heterogeneity does justice to the diversity of human driving behaviors and drives the occurrence of both “uncritical” and safety critical situations in daily traffic. Sufficient system complexity is needed to make a robust and meaningful analysis of road safety.
Challenges and Objectives
Challenge 1: A human driver model that captures the relevant behavioural mechanisms for safety assessment
- Identify causal relationships between external, and human factors, and safety-critical driver behaviors from NDS and DSE data, at the level of specific driving situations
- Make the most of unveiled patterns and existing behavioural and psychological theories to augment existing models with a 4D perception-cognitive layer
Challenge 2: Modelling the heterogeneity of human driving behaviours
- Define a methodology for generating use-cases and simulation scenarios which continuously challenge human drivers in DSE
- Map heterogeneity of human and external factors into driving performances by means of DSE, in specific driving situations
- Encode driver heterogeneity into probabilistic human behavioral models
Challenge 3: Credibility of model-based inferences
- Set up a “Modelling of the Modelling Process”
- Define a methodology to validate, or better corroborate human driver models at multiple scales
- Evaluate models in target applications
We cover the full performance spectrum of human drivers in critical driving simulations, to compare the safety performances of AVs and human-driven vehicles.
Two central ideas we propose are:
(1) a multi-level, modular and extendable simulation library that combines existing and new models for human driving behavior; in combination with
(2) an innovative cross-disciplinary methodology to account for the huge uncertainty in both human behaviors and use case circumstances.
A human driver safety baseline to i) help defining the required safety level of CCAM systems, ii) take decisions on validation requirements in type approval, iii) define fair assessment criteria in consumer testing campaigns, and iv) verify CCAM systems safety in industrial development processes. Overall i4Driving will help deploying more robust and resilient CCAM systems with a validated level of safety and security.
Developed model will serve as a reference for the automotive industry and its R&I partners to use in the design of human-like and therefore easily predictable and acceptable behaviour of ADS in mixed traffic, enabling a secure and trustworthy interaction between road users, CCAM and “conventional” vehicles, as well as safer and more efficient transport flows and a better use of infrastructure capacity.
Eventually, i4Driving models will help the automotive industry, its R&I partners, certification bodies and consumer testing organisations to realistically represent the behaviour of other human-driven vehicles in the (virtual) simulation of mixed traffic.