Search

Advances in Vehicular Cyber-Physical Systems (VCPS) are the primary enablers of the shift from no automation to fully autonomous vehicles (AVs). One of the impacts of this shift is to develop safe AVs in which most or all of the functions of the human driver are replaced with an intelligent system. However, while some progress has been made in equipping AVs with advanced AI capabilities, VCPS designers are still faced with the challenge of designing trustworthy AVs that are in sync with the unpredictable behaviours of humans. In order to address this challenge, we present a model that describes how a Human Ambassador component can be integrated into the overall design of a new generation of VCPS. A scenario is presented to demonstrate how the model can work in practice. Formalisation and co-simulation challenges associated with integrating the Human Ambassador component and future work we are undertaking are also discussed.

Driver intention recognition is essential to the development of advanced driver assistance systems providing real-time support. Current approaches for the recognition of overtaking intentions focus on drivers’ observable behaviors, neglecting the fact that the intention to overtake a slower lead car emerges earlier than the resulting behavior. This paper aims to distinguish the "intention emerging process", when drivers form the initial intention to overtake, from the "action executing process", when drivers execute the overtaking maneuver. A driving simulator study has been conducted to investigate the influence of the lead vehicle type and lead vehicle speed on initiating driver’ intention on overtaking on rural roads, and the effect of the complexity of the oncoming traffic on executing overtaking. The results show that the initial driver intention to overtake appears much earlier than the execution of the overtaking maneuver. The lead vehicle speed has a significant influence on initial driver intention in the "intention emerging process", while time to overtake increases with the number of the oncoming vehicles in the "action execution process". These results can contribute to the development of models for driver intention recognition by extending the prediction horizon from the recognition to a prediction of driving maneuvers.

Adapting the driving style of autonomous vehicles to fit the user's preferences may reduce uncomfortable user experiences. This can support the adoption process of the technology. We present ongoing research about adaptive customizable user profiles for autonomous driving styles. We propose a system that allows users to customize maneuver-based driving styles via a graphical user interface. Additionally, the system can learn user preferences based on user state assessment. We plan to design and evaluate this system with two driving simulator studies.

We performed a driving simulator study to investigate merging decisions with respect to an interaction partner in time-critical situations. The experimental paradigm was a two-alternative forced choice, where the subjects could choose to merge before human vehicles or highly automated vehicles (HAV). Under time pressure, subjects showed a significantly higher gap acceptance during merging situations when interacting with HAV. This confirmed our original hypothesis that when interacting with HAV, drivers would exploit the HAV's technological advantages and defensive programming in time-critical situations.

Highly automated vehicles, occasionally require users to resume vehicle control from non-driving related tasks by issuing cues called take-over request (TOR). Due to being engaged in non-driving related tasks (NDRT), users have a decreased level of situational awareness of the driving context. Therefore, user interface designs for TORs should ensure smooth transitions from the NDRTs to vehicle control. In this paper, we investigated the role of decision priming cues as TORs across different levels of NDRT engagement. In a driving simulator, users performed a reading span task while driving in automated mode. They received audio-visual TORs which primed them with an appropriate maneuver (steering vs. braking), depending on the traffic situation. Our results showed that priming users with upcoming maneuvers results in faster responses and longer time to collision to obstacles. However, the level of engagement in NDRT does not affect user responses to TORs.

Functional near-infrared spectroscopy (fNIRS) is a versatile imagining modality whose popularity is increasing exponentially in the neuroimaging society. Our research attempts to quantify workload in a natural driving scenario with multiple parallel tasks using fNIRS. Nine young adults participated in this study where they drove in a driving simulator for a period of 100 minutes while we continuously recorded fNIRS data. We used an n-back task to induce different workload levels forcing the participants to remember the previous one, two, three or four speed signs and adjust their speed accordingly while they interact with traffic in the virtual reality driving simulator scenario. Our results indicate that measuring the hemodynamic responses from the bilateral prefrontal cortex (PFC) can be used reliably to quantify cognitive workload levels even in more complex naturalistic tasks.

The identification of sounds is an important tool in ubiquitous and context aware applications. Today's smartphones are capable of performing even computational intensive tasks, like digital signal processing and pattern recognition. In this contribution an implementation scheme and a framework for sound recognition for smartphones are presented. A basic sound recognition flow consists of preprocessing, feature extraction, feature selection, classiication, and action trigger. A flow is not hard coded but described in a JSON file and build dynamically by the framework. The framework itself is implemented in Java for the Android operating system. But specific algorithms can be realized in Java, C(++), and Renderscript for execution on the CPU, or in Filterscript for execution on a GPU. An example flow is presented and benchmark results are shown for Java-, C-, and Filterscript-implementations of Mel Frequency Cepstral Coefficients (MFCC). Recommendations for technology selection are made.

Dose calculations with a collapsed cone algorithm implemented in a clinical treatment planning system have been studied. The algorithm has been evaluated in homogeneous as well as in heterogeneous media, and the results have been compared to measurements and Monte Carlo simulations. Commonly encountered clinical beam configurations as well as more complex geometries have been pursued to test the limitations of the model. The results show that the accuracy level reached allows for clinical use. Some situations, e.g., large wedge beams and dose calculations in the build up region, not specific to the collapsed cone model, show deviations (outside ±3%) compared to measurements. PACS number(s): 87.53.Bn, 87.53.–j, 87.66.–a

This paper presents the results of a first step in a longer - term research approach to investigate the influence of uncertainty on drivers' trust in Advanced Driver Assistance Systems (ADAS). The first step is to classify driver's uncertainty in lane changing situations. In a pilot study(n=5), the effect of distance gap on driver's uncertainty was studied using a driving simulator. The results indicated a U shaped relationship between distance gaps, reaction times and uncertainty scores.

A new three-dimensional treatment planning system (TPS) based on convolution/superposition algorithms (TMS-Radix from HELAX AB, Uppsala, Sweden) was recently installed at the University Hospital in Lund. The purpose of the present study was to design a quality assurance and acceptance testing programme to meet the specific characteristics of this convolution model. The model is based on parametrization of a non-measurable quantity-the polyenergetic pencil beam. However, the verification of the treatment planning model is still dependent on numerous comparisons of measured depth-doses and dose profiles. The test programme was divided in two basic parts: (i) model implementation and beam data consistency and (ii) model performance and limitations in special situations. The first part was scheduled for all photon beam qualities available before they could be used for clinical treatment planning. The second part was performed for selected energies only. The results indicate clearly that the model is well suited for clinical three-dimensional dose planning and that the TPS handles data as expected. For example, calculated depth-doses for open and wedge beams at depths larger than the depth of dose maximum and profiles for open beams shows a very good agreement with measurements. However, depth-dose deviations at shallow depths, especially for high energies, were found. Monitor units calculated by the system were accurate for most fields except for very large fields, where deviations of several per cent were found.