Efficient human-centered safety systems (EFFECT)

Research Project, 2023 – 2024

This project has two objectives. The first is to develop a computationally efficient advanced driver assistance systems (ADAS), in the form of an automated emergency braking system (AEB) for cars interacting with two-wheelers, based on the concept of driver comfort zone boundaries. While the algorithm should be as efficient as possible, it should also intervene as early as possible, to create larger safety margins—without creating nuisance interventions. This aim addresses three problems: 1) the computing demands of in-vehicle implementations of ADAS/AD drive system costs, 2) the far-from-perfect crash avoidance/mitigation performance of current AEB systems, and 3) the nuisance interventions that may result from addressing the second problem and intervening earlier. The second aim is to use driver monitoring to investigate the affective states of drivers during the ADAS intervention.
Understanding driver state can facilitate continuous, online improvement of ADAS (and AD) system performance by detecting and minimizing driver annoyance. Currently, ways to measure that state directly are not a part of the ADAS development process, so continuous improvement using real field data is not possible. Efficient AEB algorithms with improved performance should be easily made part of actual production systems. Although the driver-monitoring capacity is expected to reach a substantially lower TRL-level than the algorithms, it will not only facilitate long-term improvements, but also increase knowledge about the feasibility of Driver Monitoring Systems (DMS) for assessing affective state (knowledge which can be used in other applications).