Nissan Driver Knowledge and Motion SensePrincipal Investigator: David Abbink
2020-2023 (Nissan funded programme, Active)
In March 2020 a new 4-year project was started funded by Nissan. The programme funds two PhD students to study driver models for interactive traffic negotiations (Olger Siebinga, supervised by David Abbink) and motion comfort and pleasure (Rowenne Wijlens, at Aerospace, supervised by Max Mulder).
Abstract Autonomous driving (AD) technologies will greatly impact (inter)national road traffic. For the coming decade(s) traffic will consist of vehicles with different types of AD, mixed with human drivers and other road users. Current deep learning approaches lead to overly-conservative one-size-fits-all control algorithms, ill-equipped to handle mixed traffic and individual preferences safely and comfortably.
We will develop novel longitudinal and lateral control algorithms, based on a computational hierarchical controller capable of driving autonomously in a human fashion, with human-like adaptations to road conditions and traffic. We will demonstrate the controller’s capability for three key scenarios that currently are at the forefront of AD development: i) passing parked cars against oncoming traffic, ii) merging, and iii) negotiation between road users in intersections and narrow roads. Additionally, we will develop fundamental knowledge to constrain possible controller behaviour in terms of motion (dis)comfort and pleasure (what vehicle motion, feedback and controls will improve a pleasurable driving experience). We will explore this for three key scenarios that are currently already within the realm of AD: i) highway car following with wide range of speeds and accelerations, ii) gently winding roads, and iii) slowing and turning at intersection.