The great advancements in technology are transforming cars into the next digital frontier, redefining people’s lifestyle around mobility. The thesis intended to push further on this trend, exploring new interaction paradigms and creating delightful experiences in future self-driving vehicles. With a cross-discipline scope, the formula is to blend digital information into physical form and material, blurring the boundary between the car’s interior and interface. As the conclusion, I learned that a constant harmony between virtual and physical world is the key for designers to create natural and intuitive experiences with technology. The final result is an autonomous interior concept with multi-sensory user experiences. The core interface, as the physical manifestation of the car’s artificial intelligence, interact with users emotionally, offering its amazing capability in assistance. The in-car environment is evolved with sensors and displays, providing intuitive access to dedicated functions and immersive content.

This manuscript includes an investigation of decision making in alternative modes of transportation in order to understand consumers' decision in different contexts. In essay 1 of this study, the motives for participation in situated ridesharing is investigated. The study proposes a theoretical model that includes economic benefits, time benefits, transportation anxiety, trust, and reciprocity either as direct antecedents of ridesharing participation intention, or mediated through attitude towards ridesharing. Essay 2 of this study, focuses on self-driving vehicles as one of the recent innovations in transportation industry. Using a survey approach, the study develops a conceptual model of consumers' anticipated motives. Both essays use partial least square- structural equation modeling for assessing the proposed theoretical models.

With all the advancements in autonomous and connected cars, there is a developing body of research around the security and robustness of driving automation systems. Attacks and mitigations for said attacks have been explored, but almost always solely in software simulations. For this thesis, I led a team to build the foundation for an open source platoon of scale semi-autonomous vehicles. This work will enable future research into implementing theoretical attacks and mitigations. Our 1/10 scale car leverages an Nvidia Jetson, embedded microcontroller, and sensors. The Jetson manages the computer vision, networking, control logic, and overall system control; the embedded microcontroller directly controls the car. A lidar module is responsible for recording distance to the preceding car, and an inertial measurement unit records the velocity of the car itself. I wrote the software for the networking, interprocess, and serial communications, as well as the control logic and system control.

Developments of commercial self-driving vehicle (SDV) technology has a potential for a paradigm shift in traffic control technology. Contrary to some previous research approaches, this research argues that, as any other technology, traffic control technology for SDVs should be developed having in mind improved quality of life through a sustainable developmental approach. Consequently, this research emphasizes upon the social perspective of sustainability, considering its neglect in the conventional control principles, and the importance of behavioral considerations for accurately predicting impacts upon economic or environmental factors. The premise is that traffic control technology can affect the distribution of advantages and disadvantages in a society, and thus it requires a framework of social justice. The framework of social justice is inspired by John Rawls' Theory of Justice as fairness, and tries to protect the inviolability of each user in a system. Consequently, the control objective is the distribution of delay per individual, considering for example that the effect of delay is not the same if a person is traveling to a grocery store as opposed to traveling to a hospital. The notion of social justice is developed as a priority system, with end-user responsibility, where user is able to assign a specific Priority Level for each individual trip with SDV. Selected Priority Level is used to determine the right-of-way for each self-driving vehicle at an intersection. As a supporting mechanism to the priority system, there is a structure of non-monetary Priority Credits. Rules for using Priority Credits are determined using knowledge from social science research and through empirical evaluation using surveys, interviews, and web-based experiment. In the physical space, the intersection control principle is developed as hierarchical self-organization, utilizing communication, sensing, and in-vehicle technological capabilities. This distributed control approach should enable robustness against failure, and scalability for future expansion. The control mechanism has been modeled as an agent-based system, allowing evaluation of effects upon safety and user delay. In conclusion, by reaching across multiple disciplines, this development provides the promise and the challenge for evolving SDV control technology. Future efforts for SDV technology development should continue to rely upon transparent public involvement and understanding of human decision-making. / Ph. D.

The field of machine learning currently draws massive attention due to ad- vancements and successful applications announced in the last few years. One of these applications is self-driving vehicles. A machine learning model can learn to drive through behavior cloning. Behavior cloning uses an expert’s behavioral traces as training data. However, the model’s steering predictions influence the succeeding input to the model and thus the model’s input data will vary depending on earlier predictions. Eventually the vehicle may de- viate from the expert’s behavioral traces and fail due to encountering data it has not been trained on. This is the problem of sequential predictions. DAG- GER and its improvement SafeDAGGER are algorithms that enable training models in the sequential prediction domain. Both algorithms iteratively col- lect new data, aggregate new and old data and retrain models on all data to avoid catastrophically forgetting previous knowledge. The aggregation of data leads to problems with increasing model training times, memory requirements and requires that previous data is maintained forever. This thesis’s purpose is investigate whether or not SafeDAGGER can be improved with continual learning to create a more scalable and flexible algorithm. This thesis presents an improved algorithm called EWC-SD that uses the continual learning algo- rithm EWC to protect a model’s previous knowledge and thereby only train on new data. Training only on new data allows EWC-SD to have lower training times, memory requirements and avoid storing old data forever compared to the original SafeDAGGER. The different algorithms are evaluated in the con- text of self-driving vehicles on three tracks in the VBS3 simulator. The results show EWC-SD when trained on new data only does not reach the performance of SafeDAGGER. Adding a rehearsal buffer containing only 23 training exam- ples to EWC-SD allows it to outperform SafeDAGGER by reaching the same performance in half as many iterations. The conclusion is that EWC-SD with rehearsal solves the problems of increasing model training times, memory re- quirements and requiring access to all previous data imposed by data aggre- gation. / Fältet för maskininlärning drar för närvarande massiv uppmärksamhet på grund av framsteg och framgångsrika applikationer som meddelats under de senaste åren. En av dessa applikationer är självkörande fordon. En maskininlärningsmodell kan lära sig att köra ett fordon genom beteendekloning. Beteendekloning använder en experts beteendespår som träningsdata. En modells styrförutsägelser påverkar emellertid efterföljande indata till modellen och således varierar modellens indata utifrån tidigare förutsägelser. Så småningom kan fordonet avvika från expertens beteendespår och misslyckas på grund av att modellen stöter på indata som den inte har tränats på. Det här är problemet med sekventiella förutsägelser. DAGGER och dess förbättring SafeDAGGER är algoritmer som möjliggör att träna modeller i domänen sekventiella förutsägelser. Båda algoritmerna samlar iterativt nya data, aggregerar nya och gamla data och tränar om modeller på alla data för att undvika att katastrofalt glömma tidigare kunskaper. Aggregeringen av data leder till problem med ökande träningstider, ökande minneskrav och kräver att man behåller åtkomst till all tidigare data för alltid. Avhandlingens syfte är att undersöka om SafeDAGGER kan förbättras med stegvis inlärning för att skapa en mer skalbar och flexibel algoritm. Avhandlingen presenterar en förbättrad algoritm som heter EWC-SD, som använder stegvis inlärningsalgoritmen EWC för att skydda en modells tidigare kunskaper och därigenom enbart träna på nya data. Att endast träna på nya data gör det möjligt för EWC-SD att ha lägre träningstider, ökande minneskrav och undvika att lagra gamla data för evigt jämfört med den ursprungliga SafeDAGGER. De olika algoritmerna utvärderas i kontexten självkörande fordon på tre banor i VBS3-simulatorn. Resultaten visar att EWC-SD tränad enbart på nya data inte uppnår prestanda likvärdig SafeDAGGER. Ifall en lägger till en repeteringsbuffert som innehåller enbart 23 träningsexemplar till EWC-SD kan den överträffa SafeDAGGER genom att uppnå likvärdig prestanda i hälften så många iterationer. Slutsatsen är att EWC-SD med repeteringsbuffert löser problemen med ökande träningstider, ökande minneskrav samt kravet att alla tidigare data ständigt är tillgängliga som påtvingas av dataaggregering.