The introduction of more automation into our vehicles is increasing our ability to avoid or mitigate the effects of collisions. Early systems could brake when a likely collision was detected, while more advanced systems will be able to steer to avoid or reconfigure a collision during the same circumstances. This paper addresses how the post-impact motion of an impacted vehicle could be included in the decision-making process of severity minimisation motion planning. A framework is proposed that builds on previous work by the authors, combining models from motion planning, vehicle dynamics, and accident reconstruction. This framework can be configured for different contexts by adjusting its cost function according to relevant risks. Simulations of the unified system are carried out and analysed from the perspective of vehicle model complexity and collision parameters sensitivity. Additionally, effects are highlighted concerning different modelling decisions, with respect to vehicle dynamics models and collision models, that are important to consider in further research.

Full deployment of Autonomous Vehicles (AVs) on public roads is challenging for organizations in the automotive domain in terms of developing safety standards and methods while taking legacy assumptions related to having a human driver and increased complexity and complexity handling into account. Specifically, the safety of AVs in the presence of other road users must be guaranteed as far as possible for different traffic scenarios. Furthermore, unsafe situations might emerge due to uncertainty in the environment of an AV. These situations could arise due to the unexpected behaviors of others (e.g., an aggressive driver), late obstacle detection, and internal failures. Avoiding a collision with other vehicles may thus not always be possible regardless of the complexity of the planned emergency maneuver.

This thesis aims to address the problem of motion planning and control for AVs in these unique situations of unavoidable collisions. Several factors that are important in the problem formulation of a pre-crash motion planning problem for severity minimization are identified and addressed. As a result, a framework is developed that incorporates these factors and combines motion planning and control, vehicle dynamics, and accident analysis to mitigate collision risk, in particular, by reducing injury severity for vehicle occupants and increasing safety by changing the configuration of unavoidable collisions. 

This thesis tackles this problem by first proposing an algorithm that, in real-time, allows an AV to choose one action/trajectory, from a set of pre-computed trajectories, associated with the lowest injury severity for vehicle occupants. The method uses the trajectory library approach combined with numerical optimization and optimal control theories. The choice of this trajectory mainly relies upon a metric derived from accident data analysis that relates injury severity and impact location. By incorporating collision risk as a combination of collision severity and probability, the need for a configurable collision probability threshold that decides when a collision mitigation system should be activated is identified. This decision threshold balances the ability to reduce collision severity with the undesired increase in the likelihood of a collision taking place.

The studies included in this thesis show that different decision-making strategies involving decision thresholds for collision mitigation/reconfiguration systems can lead to statistically significant differences in the resulting collision severity. Furthermore, unobserved heterogeneity may arise through the introduction of these systems, e.g., due to slight variations in the parameters of the algorithms they employ. The problem of motion planning in unavoidable collisions is further extended by proposing a unified system that incorporates the risks of post-impact motions resulting from the original impact. The extended framework can be configured for different contexts by adjusting its cost function according to relevant post-impact risks. 

The result of this thesis aims to contribute to the field of motion planning in unavoidable collisions and to provide guidance for further improvement of road safety. Further research is required to fully explore this field and address the challenges of motion planning and control in unavoidable collision scenarios.

Att förverkliga kommersiell drift av autonoma vägfordon (AV) på allmänna vägar är en utmaning för organisationer i vägfordonsdomänen i fråga om att utveckla personsäkerhetsstandarder och metoder samtidigt som kvardröjande antaganden relaterade till existensen av en mänsklig förare, och ökad komplexitet och hantering av komplexitet, beaktas. Det som gör utmaningarna speciellt svåra att hantera är att AV i största möjliga mån måste kunna garantera att de inte påverkar personsäkerheten för andra trafikanter negativt i olika trafikscenarios. Dessutom, trots dessa garantier kan farliga situationer uppstå på grund av osäkerheten i miljön som AV rör sig i. Dessa situationer kan uppstå på grund av oförutsett beteende från andra (t.ex. en aggressiv förare), sen upptäckt av hinder, och interna fel. Att undvika en kollision med andra fordon är därför inte alltid möjligt oavsett komplexiteten i den planerade akuta undanmanövern.

Den här avhandlingen syftar till att adressera problembilden runt rörelse-planering och -kontroll av AV i dessa unika situationer där en kollision är omöjlig att undvika. Åtskilliga faktorer som är viktiga för formuleringen av en problembild för rörelseplanering, som utförs innan en oundviklig kollision och är ämnad att reducera omfattningen av skador, identifieras och adresseras. Detta har resulterat i utvecklandet av ett ramverk som integrerar dessa faktorer och kombinerar rörelse-planering och -kontroll, fordonsdynamik, och analys av olyckor för att mildra kollisionsrisk, specifikt genom att reducera skadeomfattningen för passagerare i fordon, och öka personsäkerheten genom att ändra konfigureringen av oundvikliga kollisioner.

Den här avhandlingen föreslår att problembilden hanteras genom en algoritm som, i realtid, låter en AV välja den handling/bana, från flera föruträknade manövrar, som kan associeras med den minsta skadeomfattningen för fordonspassagerare. Metoden använder sig av ett angreppssätt som kombinerar ett bibliotek av banor med numerisk optimering och optimal kontrollteori. Valet av bana baseras i huvudsak på ett mått härlett ur analys av data från olyckor som relaterar skadeomfattning till var på fordonet en kollision inträffar. Genom att inkludera kollisionsrisk som en kombination av en kollisions skadeomfattning och sannolikhet, identifieras behovet av en konfigurerbar tröskel för kollisionssannolikhet som kan avgöra när ett system för att mildra effekterna av kollisioner bör aktiveras. Denna tröskel balanserar möjligheten att reducera omfattningen av skador orsakade av kollisioner med en relaterad, oönskad ökning av sannolikheten att kollisioner inträffar.

Studierna som lagts samman i den här avhandlingen visar att olika strategier för beslutsfattande, när en tröskel för kollisionssannolikhet används i ett system för att mildra effekterna av kollisioner, kan leda till statistiskt signifikanta skillnader i omfattningen av de skador som orsakas av kollisioner. Oobserverad heterogenitet kan dessutom uppstå när dessa system introduceras, t.ex. på grund av små variationer av parametrarna i algoritmerna de använder sig av. Problembilden runt rörelseplanering när en kollision inte kan undvikas utökas ytterligare genom att ett enhetligt system föreslås, vilket även kan hantera risker relaterade till de rörelser som kommer efter och beror på denna initiala kollision. Det utökade ramverket kan konfigureras för olika kontexter genom att dess kostnadsfunktion justeras utifrån de risker som är relevanta efter en initial kollision.

Resultaten som presenteras i den här avhandlingen syftar till att bidra till forskningsfältet runt rörelseplanering innan oundvikliga kollisioner och tillhandahålla vägledning för ytterligare förbättring av vägsäkerhet. Ytterligare forskning behövs för att helt utforska det här forskningsfältet och adressera utmaningarna för rörelse-planering och -kontroll i scenarier som involverar oundvikliga kollisioner.

The deployment of autonomous vehicles on public roads calls for the development of methods that are reliably able to mitigate injury severity in case of unavoidable collisions. This study proposes a data-driven motion planning method capable of minimizing injury severity for vehicle occupants in unavoidable collisions. The method is based on establishing a metric that models the relationship between impact location and injury severity using real accident data, and subsequently including it in the cost function of a motion planning framework. The vehicle dynamics and associated constraints are considered through a precomputed trajectory library, which is generated by solving an optimal control problem. This allows for efficient computation as well as an accurate representation of the vehicle. The proposed motion planning approach is evaluated by simulation, and it is shown that the trajectory associated with the minimum cost mitigates the collision severity for occupants of passenger vehicles involved in the collision.

We present a probabilistic collision mitigation system (CMS) that aims to decrease the severity of unavoidable collisions. CMSs have to decide when to act in order to keep enough trajectories for lowering severity available, while at the same time not increasing the collision probability. This paper operationalises the choice of when to act as a probability threshold, which is applied to all or part of the trajectories available to an automated vehicle equipped with the CMS. Through simulations the relationship between the value of the threshold, and the trade-off between being able to reduce the collision severity and not causing a collision, is investigated. We identify two types of behaviours that increase the severity of collisions: one when acting late (setting a higher threshold) reduces the number of available trajectories with lower severity, and one when acting early is susceptible to prediction errors. Regardless, by setting a lower prediction horizon the prediction accuracy increases, but also reduces the available trajectories with lower severity. We also note that the outcome of the CMS improves when the severity of a trajectory is aligned with maximum risk of collision along it, rather than the severity at the time step when it crosses the probability threshold.

This paper addresses the control of a highly automated vehicle in a traffic scenario, where colliding with other traffic agents is unavoidable. Such a critical situation could be the result of a fault in the vehicle, late obstacle detection or the presence of an aggressive driver. We provide an approach that allows the vehicle’s control system to choose the manoeuvre that is likely to lead to the least severe injuries to vehicle occupants.The approach involves the off-line solving of an optimal control problem to create a set of trajectories based on controlling the steering angle rate and the braking rate at the vehicle’s limits. Occupant injury severity prediction, based on accident data with the focus on impact location, is used by a real-time collision control algorithm to choose a trajectory from the pre-computed optimal set. A simulation set-up is presented to illustrate the idea of the collision control algorithm in a simple scenario involving dynamic traffic agents.

Organizations in the automotive domain, which aim to transition into developing fully autonomous vehicles face many challenges. These range from organizational issues to engineering concerns. This paper builds on structured interviews with professionals from industry and academia to provide a deeper understanding of existing problems. Standards, safety analysis, legacy assumptions related to having a human driver, and increased complexity and complexity handling were raised as important concerns. The analysis of these concern leads us to consider the current relationship between academia and industry as too disconnected. There is a risk that new techniques developed by academia end up irrelevant to industry. This underlying problem, and others relevant to autonomy, might be solved by collaborative research between different automotive companies. However, there are experts that challenge the underlying need for such collaboration. Therefore, externally to automotive companies, new expert arenas might be required in order to facilitate an exchange of ideas that lead to new collaboration efforts. Internally to automotive companies, the changes brought on by autonomy will lead to organizational changes and the creation of new roles. These organizational changes will have to be managed, or otherwise unnecessary conflict might occur between new and old roles.

The introduction of more automation into our vehicles is increasing our ability to avoid or mitigate the effects of collisions. Early systems could brake when a likely collision was detected, while more advanced systems will be able to steer to avoid or reconfigure a collision during the same circumstances. This paper addresses how the post-impact motion of an impacted vehicle could be included in the decision making process of severity minimization motion planning. A framework is proposed that builds on previous work by the authors, combining models from motion planning, vehicle dynamics and accident reconstruction. This framework can be configured for different contexts by adjusting its cost function according to relevant risks. Simulations of the unified system are carried out and analysed from the perspective of vehicle model complexity and collision parameters sensitivity. Additionally, effects are highlighted concerning different modeling decisions, with respect to vehicle dynamics models and collision models, that are important to consider in further research.  

Autonomous vehicles are equipped with advanced vehicle technology (AVT) that will improve road traffic safety and reduce accidents. However, due to the uncertain behavior of other road users, collisions can never be completely eliminated. Collision reconfiguration systems offer a solution by, for instance, changing where vehicles are hit and how the impact force is directed towards them. Unfortunately, the logic behind the decision-making of collision reconfiguration systems is fundamentally different from that of other AVTs. Fundamentally different feedback might thus be required from accident analyses to ensure the successful design of collision reconfiguration systems. Through simulations, this study explores decision-making strategies of collision reconfiguration systems to ascertain the implications of which feedback is required from accident analyses. Results show that different strategies can be statistically significantly different from each other in the way they affect severity; and that a new source of unobserved heterogeneity could easily be small variations in the algorithms used by collision reconfiguration systems. Based on this, three new needs to consider during accident analysis are put forth: firstly, new safety surrogate measures (SSMs) that consider severity are required; one such SSM is proposed; secondly, to identify new unobserved heterogeneity as a result of collision reconfiguration systems, the trajectories of traffic near-collisions should be recorded, and statistical tools to identify comparable scenarios developed. Thirdly, new collision patterns will make it difficult to analyze the implications of reconfigured collisions, which suggests that collision configurations must be carefully recorded to provide early feedback.