abstract: As robots become mechanically more capable, they are going to be more and more integrated into our daily lives. Over time, human’s expectation of what the robot capabilities are is getting higher. Therefore, it can be conjectured that often robots will not act as human commanders intended them to do. That is, the users of the robots may have a different point of view from the one the robots do. The first part of this dissertation covers methods that resolve some instances of this mismatch when the mission requirements are expressed in Linear Temporal Logic (LTL) for handling coverage, sequencing, conditions and avoidance. That is, the following general questions are addressed: * What cause of the given mission is unrealizable? * Is there any other feasible mission that is close to the given one? In order to answer these questions, the LTL Revision Problem is applied and it is formulated as a graph search problem. It is shown that in general the problem is NP-Complete. Hence, it is proved that the heuristic algorihtm has 2-approximation bound in some cases. This problem, then, is extended to two different versions: one is for the weighted transition system and another is for the specification under quantitative preference. Next, a follow up question is addressed: * How can an LTL specified mission be scaled up to multiple robots operating in confined environments? The Cooperative Multi-agent Planning Problem is addressed by borrowing a technique from cooperative pathfinding problems in discrete grid environments. Since centralized planning for multi-robot systems is computationally challenging and easily results in state space explosion, a distributed planning approach is provided through agent coupling and de-coupling. In addition, in order to make such robot missions work in the real world, robots should take actions in the continuous physical world. Hence, in the second part of this thesis, the resulting motion planning problems is addressed for non-holonomic robots. That is, it is devoted to autonomous vehicles’ motion planning in challenging environments such as rural, semi-structured roads. This planning problem is solved with an on-the-fly hierarchical approach, using a pre-computed lattice planner. It is also proved that the proposed algorithm guarantees resolution-completeness in such demanding environments. Finally, possible extensions are discussed. / Dissertation/Thesis / Doctoral Dissertation Computer Science 2019

Intelligent transportation systems involve a variety of information and control systems methodologies, from cooperative systems which aim at traffic flow optimization by means of vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communication, to information fusion from multiple traffic sensing modalities. This thesis aims to address three problems in intelligent transportation systems, one in optimal control of connected automated vehicles, one in discrete-event and hybrid traffic simulation model, and one in sensing and classifying roadway obstacles in smart cities. The first set of problems addressed relates to optimally controlling connected automated vehicles (CAVs) crossing an urban intersection without any explicit traffic signaling. A decentralized optimal control framework is established whereby, under proper coordination among CAVs, each CAV can jointly minimize its energy consumption and travel time subject to hard safety constraints. A closed-form analytical solution is derived while taking speed, control, and safety constraints into consideration. The analytical solution of each such problem, when it exists, yields the optimal CAV acceleration/deceleration. The framework is capable of accommodating for turns and ensures the absence of collisions. In the meantime, a measurement of passenger comfort is taken into account while the vehicles make turns. In addition to the first-in-first-out (FIFO) ordering structure, the concept of dynamic resequencing is introduced which aims at further increasing the traffic throughput. This thesis also studies the impact of CAVs and shows the benefit that can be achieved by incorporating CAVs to conventional traffic. To validate the effectiveness of the proposed solution, a discrete-event and hybrid simulation framework based on SimEvents is proposed, which facilitates safety and performance evaluation of an intelligent transportation system. The traffic simulation model enables traffic study at the microscopic level, including new control algorithms for CAVs under different traffic scenarios, the event-driven aspects of transportation systems, and the effects of communication delays. The framework spans multiple toolboxes including MATLAB, Simulink, and SimEvents. In another direction, an unsupervised anomaly detection system is developed based on data collected through the Street Bump smartphone application. The system, which is built based on signal processing techniques and the concept of information entropy, is capable of generating a prioritized list of roadway obstacles, such that the higher-ranked entries are most likely to be actionable bumps (e.g., potholes) requiring immediate attention, while those lower-ranked are most likely to be nonactionable bumps(e.g., flat castings, cobblestone streets, speed bumps) for which no immediate action is needed. This system enables the City to efficiently prioritize repairs. Results on an actual data set provided by the City of Boston illustrate the feasibility and effectiveness of the system in practice.

This thesis investigates the design of a local planning method for a reversing single joint tractor-trailer system that can be used in a sampling-based motion planner. The motion planner used is a Rapidly-exploring Random Tree (RRT) developed by Scania. The main objective of a local planning method is to generate a feasible path between two poses, which is needed when expanding the search tree in an RRT. The local planning method described in this thesis uses a set of curves, similar to Reeds-Shepp curves, feasible for a single joint tractor-trailer system. The curves are found by solving a constrained optimization problem that adheres to the kinematic model of the system. The reference for the tractor is generated by discretizing the path between curves. The reference for the trailer is generated by simulating the mission backwards where the curve radiuses are used as input. Simulating the mission backwards circumvents the instability of the system when reversing. The generated references are then compared to references generated by a lattice-based motion planner. The length of the references generated by the RRT are smaller than those generated by the lattice-based motion planner in simple open environments. The RRT had issues finding a path in cases where the environment was complex while the lattice-based motion planner found a path in every scenario. The computational time was significantly lower for the RRT in all simulations. The RRT generates all references between any two given poses while the lattice-based motion planner approximate the start and goal poses to the closest vertex in the search-space.  The references generated by the RRT did not perform optimally when small turns were needed along the curves due to the distance needed for maneuvering the tractor. Therefore, a new optimization problem has to be defined for which the small turns are considered.

Cette thèse présente une approche de modélisation inspirée par la robotique pour l'étude des changements conformationnels des protéines. Cette approche est basée sur une représentation mécanistique des protéines permettant l'application de méthodes efficaces provenant du domaine de la robotique. Elle fournit également une méthode appropriée pour le traitement « gros-grains » des protéines sans perte de détail au niveau atomique. L'approche présentée dans cette thèse est appliquée à deux types de problèmes de simulation moléculaire. Dans le premier, cette approche est utilisée pour améliorer l'échantillonnage de l'espace conformationnel des protéines. Plus précisément, cette approche de modélisation est utilisée pour implémenter des classes de mouvements pour l'échantillonnage, aussi bien connues que nouvelles, ainsi qu'une stratégie d'échantillonnage mixte, dans le contexte de la méthode de Monte Carlo. Les résultats des simulations effectuées sur des protéines ayant des topologies différentes montrent que cette stratégie améliore l'échantillonnage, sans toutefois nécessiter de ressources de calcul supplémentaires. Dans le deuxième type de problèmes abordés ici, l'approche de modélisation mécanistique est utilisée pour implémenter une méthode inspirée par la robotique et appliquée à la simulation de mouvements de grande amplitude dans les protéines. Cette méthode est basée sur la combinaison de l'algorithme RRT (Rapidly-exploring Random Tree) avec l'analyse en modes normaux, qui permet une exploration efficace des espaces de dimension élevée tels les espaces conformationnels des protéines. Les résultats de simulations effectuées sur un ensemble de protéines montrent l'efficacité de la méthode proposée pour l'étude des transitions conformationnelles / Proteins are biological macromolecules that play essential roles in living organisms. Un- derstanding the relationship between protein structure, dynamics and function is indis- pensable for advances in fields such as biology, pharmacology and biotechnology. Study- ing this relationship requires a combination of experimental and computational methods, whose development is the object of very active interdisciplinary research. In such a context, this thesis presents a robotics-inspired modeling approach for studying confor- mational changes in proteins. This approach is based on a mechanistic representation of proteins that enables the application of efficient methods originating from the field of robotics. It also provides an accurate method for coarse-grained treatment of proteins without loosing full-atom details.The presented approach is applied in this thesis to two different molecular simulation problems. First, the approach is used to enhance sampling of the conformational space of proteins using the Monte Carlo method. The modeling approach is used to implement new and known Monte Carlo trial move classes as well as a mixed sampling strategy. Results of simulations performed on proteins with different topologies show that this strategy enhances sampling without demanding higher computational resources. In the second problem tackled in this thesis, the mechanistic modeling approach is used to implement a robotics-inspired method for simulating large amplitude motions in proteins. This method is based on the combination of the Rapidly-exploring Random Tree (RRT) algorithm with Normal Mode Analysis (NMA), which allows efficient exploration of the high dimensional conformational spaces of proteins. Results of simulations performed on ten different proteins of different sizes and topologies show the effectiveness of the proposed method for studying conformational transitions

Un robot agit sur son environnement par le mouvement, sa capacité à planifier ses mouvements est donc une composante essentielle de son autonomie. L'objectif de cette thèse est concevoir des méthodes algorithmiques performantes permettant le calcul automatique de trajectoires pour des systèmes robotiques complexes dans le cadre de la robotique d'assistance. Les systèmes considérés qui ont pour vocation de servir l'homme et de l'accompagner dans des tâches du quotidien doivent tenir compte de la sécurité et du bien-être de l'homme. Pour cela, les mouvements du robot doivent être générés en considérant explicitement le partenaire humain raisonant sur un modèle du comportement social de l'homme, de ses capacités et de ses limites afin de produire un comportement synergique optimal.Dans cette thèse nous étendons les travaux pionniers menés au LAAS dans ce domaine afin de produire des mouvements considérant l’homme de manière explicite dans des environnements encombrés. Des algorithmes d’exploration de l’espace des configurations par échantillonnage aléatoire sont combinés à des algorithmes d’optimisation de trajectoire afin de produire des mouvements sûrs et agréables. Nous proposons dans un deuxième temps un planificateur de tâche d’échange d’objet prenant en compte la mobilité du receveur humain permettant ainsi de partager l’effort lors du transfert. La pertinence de cette approche a été étudiée dans une étude utilisateur. Finalement, nous présentons une architecture logicielle qui permet de prendre en compte l’homme de manière dynamique lors de la réalisation de tâches de manipulation interactive / A robot act upon its environment through motion, the ability to plan its movements is therefore an essential component of its autonomy. The objective of this thesis is to design algorithmic methods to perform automatic trajectory computation for complex robotic systems in the context of assistive robotics. This emerging field of autonomous robotics applications brings new constraints and new challenges. Such systems that are designed to serve humans and to help in daily tasks must consider the safety and well-being of the surrounding humans. To do this, the robot's motions must be generated by considering the human partner explicitly. For comfort and efficiency, the robot must take into account a model of human social behavior, capabilities and limitations to produce an optimal synergistic behavior.In this thesis we extend to cluttered environments the pioneering work that has been conducted at LAAS in this field. Algorithms that explore the configuration space by random sampling are combined with trajectory optimization algorithms to produce safe and human aware motions. Secondly we propose a planner for object handover taking into account the mobility of the human recipient allowing to share the effort during the transfer. The relevance of this approach has been studied in a user study. Finally, we present a software architecture developed in collaboration with a partner of the European project Dexmart that allows to take dynamically into account humans during the execution of interactive manipulation tasks

Le problème de la planification du mouvement a été largement étudié dans le cas de corps rigides articulés mais peu de travaux considèrent les corps déformables. En particulier, les tiges élastiques telles que cables électriques, flexibles hydrauliques et pneumatiques, apparaissent dans de nombreux contextes industriels. Du fait d'une modélisation complexe et d'un grand nombre de degrés de liberté, l'extension des méthodes de planification de mouvement à de tels corps est un problème particulièrement difficile. En se basant sur les propriétés des configurations à l'équilibre statique, cette thèse propose plusieurs approches au problème de planification de mouvement pour des tiges élastiques. / The motion planning problem has been broadly studied in the case of articulated rigid body systems but so far few work have considered deformable bodies. In particular, elastic rods such as electric cables, hydraulic or pneumatic hoses, appear in many industrial contexts. Due to complex models and high number of degrees of freedom, the extension of motion planning methods to such bodies is a difficult problem. By taking advantage of the properties of static equilibrium configurations, this thesis presents several approaches to the motion planning problem for elastic rods.

La planification de mouvement est le domaine de l’informatique qui a trait au développement de techniques algorithmiques permettant la génération automatique de trajectoires pour un système mécanique. La nature d’un tel système varie selon les champs d’application. En animation par ordinateur il peut s’agir d’un avatar humanoïde. En biologie moléculaire cela peut être une protéine. Le domaine d’application de ces travaux étant la robotique aérienne, le système est ici un UAV (Unmanned Aerial Vehicle: véhicule aérien sans pilote) à quatre hélices appelé quadrirotor. Le problème de planification de mouvements consiste à calculer une série de mouvements qui amène le système d’une configuration initiale donnée à une configuration finale souhaitée sans générer de collisions avec son environnement, la plupart du temps connu à l’avance. Les méthodes habituelles explorent l’espace des configurations du système sans tenir compte de sa dynamique. Cependant, la force de poussée qui permet à un quadrirotor de voler est par construction parallèle aux axes de rotation des hélices, ce qui implique que certains mouvements ne peuvent pas être effectués. De plus, l’intensité de cette force de poussée, et donc l’accélération linéaire du centre de masse, sont limitées par les capacités physiques du robot. Pour toutes ces raisons, non seulement la position et l’orientation doivent être planifiées, mais les dérivées plus élevées doivent l’être également si l’on veut que le système physique soit en mesure de réellement exécuter le mouvement. Lorsque c’est le cas, on parle de planification kinodynamique de mouvements. Une distinction est faite entre le planificateur local et le planificateur global. Le premier est chargé de produire une trajectoire valide entre deux états du système sans nécessairement tenir compte des collisions. Le second est l’algorithme principal qui est chargé de résoudre le problème de planification de mouvement en explorant l’espace d’état du système. Il fait appel au planificateur local. Nous présentons un planificateur local qui interpole deux états comprenant un nombre arbitraire de degrés de liberté ainsi que leurs dérivées premières et secondes. Compte tenu d’un ensemble de limites sur les dérivées des degrés de liberté jusqu’au quatrième ordre (snap), il produit rapidement une trajectoire en temps minimal quasi optimale qui respecte ces limites. Dans la plupart des algorithmes modernes de planification de mouvements, l’exploration est guidée par une fonction de distance (ou métrique). Le meilleur choix pour celle-ci est le cost-to-go, c.a.d. le coût associé à la méthode locale. Dans le contexte de la planification kinodynamique de mouvements, il correspond à la durée de la trajectoire en temps minimal. Le problème dans ce cas est que calculer le cost-to-go est aussi difficile (et donc aussi coûteux) que de calculer la trajectoire optimale elle-même. Nous présentons une métrique qui est une bonne approximation du cost-to-go, mais dont le calcul est beaucoup moins coûteux. Le paradigme dominant en planification de mouvements aujourd’hui est l’échantillonnage aléatoire. Cette classe d’algorithmes repose sur un échantillonnage aléatoire de l’espace d’état afin de l’explorer rapidement. Une stratégie commune est l’échantillonnage uniforme. Il semble toutefois que, dans notre contexte, ce soit un choix assez médiocre. En effet, une grande majorité des états uniformément échantillonnés ne peuvent pas être interpolés. Nous présentons une stratégie d’échantillonnage incrémentale qui diminue considérablement la probabilité que cela ne se produise. / Motion planning is the field of computer science that aims at developing algorithmic techniques allowing the automatic computation of trajecto- ries for a mechanical system. The nature of such a system vary according to the fields of application. In computer animation it could be a humanoid avatar. In molecular biology it could be a protein. The field of application of this work being aerial robotics, the system is here a four-rotor UAV (Unmanned Aerial Vehicle) called quadrotor. The motion planning problem consists in computing a series of motions that brings the system from a given initial configuration to a desired final configuration without generating collisions with its environment, most of the time known in advance. Usual methods explore the system’s configuration space regardless of its dynamics. By construction the thrust force that allows a quadrotor to fly is tangential to its attitude which implies that not every motion can be performed. Furthermore, the magnitude of this thrust force and hence the linear acceleration of the center of mass are limited by the physical capabilities of the robot. For all these reasons, not only position and orientation must be planned, higher derivatives must be planned also if the motion is to be executed. When this is the case we talk of kinodynamic motion planning. A distinction is made between the local planner and the global planner. The former is in charge of producing a valid trajectory between two states of the system without necessarily taking collisions into account. The later is the overall algorithmic process that is in charge of solving the motion planning problem by exploring the state space of the system. It relies on multiple calls to the local planner. We present a local planner that interpolates two states consisting of an arbitrary number of degrees of freedom (dof) and their first and second derivatives. Given a set of bounds on the dof derivatives up to the fourth order (snap), it quickly produces a near-optimal minimum time trajectory that respects those bounds. In most of modern global motion planning algorithms, the exploration is guided by a distance function (or metric). The best choice is the cost-to-go, i.e. the cost associated to the local method. In the context of kinodynamic motion planning, it is the duration of the minimal-time trajectory. The problem in this case is that computing the cost-to-go is as hard (and thus as costly) as computing the optimal trajectory itself. We present a metric that is a good approximation of the cost-to-go but which computation is far less time consuming. The dominant paradigm nowadays is sampling-based motion planning. This class of algorithms relies on random sampling of the state space in order to quickly explore it. A common strategy is uniform sampling. It however appears that, in our context, it is a rather poor choice. Indeed, a great majority of uniformly sampled states cannot be interpolated. We present an incremental sampling strategy that significantly decreases the probability of this happening.

One of the biggest challenges associated with optimization based methods forrobotic motion planning is their extreme sensitivity to a good initial guess,especially in the presence of local minima in the cost function landscape.Additional challenges may also arise due to operational constraints, robotcontrollers sometimes have very little time to plan a trajectory to perform adesired function. To work around these limitations, a common solution is tosplit the motion planner into an offline phase and an online phase. The offlinephase entails computing reference trajectories for varying parameterizationsof the task space in the form of a lookup table. During the online phase,a stripped down version of the optimizer is supplied with a suitable initialguess from the lookup table using the current state estimate of the robot andits surrounding bodies. This method helps in alleviating problems related toboth local minima and operational time constraints, by seeding the optimizerwith a suitable initial guess that allows it to converge to the global minimummuch faster.The problem however, shifts to the computational complexity of computinga lookup table of reference trajectories for a fine enough discreti- zation ofthe input state space. For many robotic scenarios of interest, it is oftenimpractical and sometimes computationally infeasible to compute a look uptable using a serial, single core implementation of the offline phase of a motionplanner. The main contribution of this work is to develop and evaluate amethod for reducing the time spent on computing a lookup table of referencetrajectories during the offline phase of motion planners based on optimalcontrol. We implement a method to offload the computation of collisionavoidance constraints during trajectory optimization on a Graphics ProcessingUnit (GPU), while simultaneously benefiting from a task based approach todistribute lookup table computations for independent subsets of the input statespace across multiple processes on a cluster of machines. We demonstrate theefficacy of the proposed method in a practical setting by implementing andevaluating it within a representative motion planner based on optimal control.We observe that the implemented method is 115x faster than the originalserial version of the planner, using 86 processes on 5 machines with standardserver grade hardware and 5 Graphics Processing Units in total. Additionally,we observe that the implemented method results in solutions identical to theoriginal serial version in 96.6% of cases, lending credibility for its use inrobotic motion planning. / En av de största utmaningarna med optimeringsbaserade metoder för rörelseplaneringinom robotik är deras extrema känslighet för en bra initial gissning,särskilt i närvaro av lokala minima i kostnadsfunktionslandskapet. Ytterligareutmaningar kan också uppstå på grund av operativa begränsningar. Robotkontrollerhar ibland väldigt lite tid att planera en väg för att utföra en önskadfunktion. För att kringgå dessa begränsningar är en vanlig lösning att dela upprörelseplaneraren i en offline-fas och en online-fas. Offlinefasen inkluderarberäkning av referensvägar för olika punkter i ingångstillståndsutrymmet iform av en uppslagstabell. Under online-fasen levereras en avskalad versionav optimeraren med en lämplig initial gissning från uppslagstabellen medden aktuella uppskattningen av roboten och dess omgivande kroppar. Dennametod hjälper till att lindra problem relaterade till både lokala minima ochdriftstidsbegränsningar genom att sådd optimeraren med en lämplig initialgissning som gör att den kan konvergera till det globala minimumet mycketsnabbare.Problemet flyttas emellertid nu till beräkningskomplexiteten för att beräknaen uppslagstabell över referensvägar för ett tillräckligt fint utrymme för ingångstillståndsutrymmet.För många robotscenarier av intresse är det ofta opraktisktoch ibland beräkningsmässigt omöjligt att beräkna en uppslagstabell med hjälpav en seriell, enda kärnimplementering av offline-fasen i en rörelseplanner.Huvudbidraget till detta arbete är att utveckla och utvärdera en metod för attminska tiden som används för att beräkna en uppslagstabell över referensvägarunder offline-fasen för rörelsesplanerare baserat på optimal kontroll. Vi implementeraren metod för att utföra en kollision undvika en grafikbehandlingsenhet(GPU), medan du använder en uppgiftsbaserad metod för att distribuerauppslagningsberäkningar för oberoende delmängder av inmatningsutrymmeöver flera processer i ett kluster av maskiner. Vi demonstrerar effektivitetenav den föreslagna metoden i en praktisk miljö genom att implementeraoch utvärdera den inom en representativ rörelseplanner baserat på optimalkontroll. Vi noterar att den implementerade metoden är 115 gånger snabbareän den ursprungliga serieversionen av schemaläggaren, med 86 processer på 5maskiner med standardhårdvara och totalt 5 GPU: er. Dessutom observerarvi att den implementerade metoden resulterar i lösningar som är identiskamed den ursprungliga serieversionen i mer än 96,6 % av fallen, vilket gertrovärdighet för dess användning i robotrörelse planering.

Many of the applications that have been proposed for future small unmanned aerial vehicles (UAVs) are at low altitude in areas with many obstacles. A vital component for successful navigation in such environments is a path planner that can find collision free paths for the UAV. Two popular path planning algorithms are the probabilistic roadmap algorithm (PRM) and the rapidly-exploring random tree algorithm (RRT). Adaptations of these algorithms to an unmanned autonomous helicopter are presented in this thesis, together with a number of extensions for handling constraints at different stages of the planning process. The result of this work is twofold: First, the described planners and extensions have been implemented and integrated into the software architecture of a UAV. A number of flight tests with these algorithms have been performed on a physical helicopter and the results from some of them are presented in this thesis. Second, an empirical study has been conducted, comparing the performance of the different algorithms and extensions in this planning domain. It is shown that with the environment known in advance, the PRM algorithm generally performs better than the RRT algorithm due to its precompiled roadmaps, but that the latter is also usable as long as the environment is not too complex. The study also shows that simple geometric constraints can be added in the runtime phase of the PRM algorithm, without a big impact on performance. It is also shown that postponing the motion constraints to the runtime phase can improve the performance of the planner in some cases. / <p>Report code: LiU–Tek–Lic–2006:10.</p>

The present study proposes a novel approach to Coverage Path Planning for unmanned aerial vehicle (UAV) inspired by the Orienteering Problem. The main goal is to develop an algorithm suitable for Search and Rescue Missions, which can produce a search pattern with dynamical constrains, that is not limited to the traditional back-and-forth motion or spiral patterns. This method leads to a more flexible and diverse coverage of the Area of Interest. In order to generate dynamically correct trajectories, we utilize cubic splines as motion primitives to solve the Orienteering Problem. To accomplish this, we implement and test three different types of cubic splines, namely Catmull-Rom, Freya, and B-splines. To determine the coverage of the search area, the sensor's projection (footprint) is evaluated along the spline trajectory onto a costmap. This method accounts for the footprint's orientation and size, which depend on the UAV's attitude to some extent. This version of the Orienteering Problem using splines for dynamical control and calculating coverage, we call the Mapping Motion Orienteering Problem (MMOP). \\The heuristic method Simulated Annealing is used to address the combinatorial challenges of the MMOP, and two cost functions are tested for optimization. The study shows that the choice of spline has a significant impact on the algorithm's efficacy, and B-splines are the most effective in generating dynamic and adaptable trajectories. However, the study also shows that the Simulated Annealing algorithm with identical settings produced varied resulting paths. Finally, further research is needed to solve the challenges faced with the computational time, which heavily depends on factors such as the sampling rate for the footprint along the path and the resolution of the costmap and footprint itself.