Hydropower is an energy source that is currently considered to be both climate- and environmentally friendly. It is utilized on both large and small scales and has a low carbon footprint, which has led to increased attention in recent years. The growing influence of renewable energy sources such as solar and wind power has also brought the regulatory capabilities of hydropower within electrical grids into focus. Despite the significant climate benefits of hydropower, there remain substantial challenges in adapting it to minimize environmental impacts. The latest EU directive has highlighted how fish and other aquatic organisms are harmed by limitations on upstream and downstream migration caused by hydropower plants.

Given the importance of maximizing hydropower electricity production while minimizing environmental impacts, there is a need for more knowledge regarding how mitigation measures for aquatic organisms can be implemented. By combining knowledge of water flow patterns in hydropower areas with insights into the behavior of fish species in these environments, it is possible to facilitate migration both upstream and downstream while ensuring optimal use of hydropower.

The first section of this thesis focuses on downstream-migrating salmon smolts. In this study, telemetry data from tagged smolts were analyzed alongside a 3D flow model of an area in northern Sweden where a large hydropower plant affects one of the largest rivers for salmon reproduction. The study shows that the smolts follow the main channel and are influenced by the flow rate. Higher flow velocities tend to cause the smolts to concentrate more in the main channel, whereas lower flows result in a broader distribution across the riverbed. The smolts are also partially influenced by the boom installed to direct the flow towards the fish pass. By studying telemetry tracks and CFD in detail, one objective has been to develop a method for integrating these two types of data, thereby creating a behavioral model of how smolt navigate at different flow velocities.

The second work section addresses how climate change may impact flow events in a dammed river in northern Sweden. With anticipated climate change, more significant variation in precipitation is expected to affect the northern hemisphere, resulting in altered flow conditions. By studying historical data, an extreme flow event was identified and modeled using a 2D model. Flow variations were analyzed in relation to the preferred flow conditions for grayling. The study demonstrates that grayling are sensitive to large flow variations in the area. The most significant impact occurs when an extreme flow coincides with their spawning period. As the area is heavily regulated with hydropower plants placed closely together, downstream plants also have an impact on water levels upstream. The area is particularly sensitive if water levels are rapidly reduced, as this can lead to stranding. Since hydropeaking is a factor that affects the aquatic environment and is expected to become more pronounced in the future, it has also been studied in greater detail. By modeling how the river reach is influenced by different dewatering periods, it has been possible to identify specific types of zero-flow events that are particularly critical. Based on these findings, various mitigation measures have been proposed to guide hydropower operators in planning peak generation to reduce environmental impacts while accounting for potential energy production losses.

This thesis combines various methods to investigate flow and flow variations, utilizing both detailed 3D models and broader 2D models. It demonstrates the potential for flow models to interact with ecological studies to deepen our understanding of the ecological state of rivers affected by hydropower development. 

Agricultural biomass is increasingly acknowledged as a versatile renewable feedstock in the energy conversion units. However, the efficient utilization of agricultural biomass in thermochemical processes could be hindered by the relatively high ash content compared to woody biomass. These types of biomasses often contain a relatively high share of silicon (Si) and phosphorus (P). The presence of these elements in the biomass can contribute to ash-related operational challenges such as slagging, fine particle emissions, and deposit formation. Beyond these challenges, the recovery of Si- and P-containing compounds as by-products during thermochemical conversion presents an opportunity to generate additional value, thereby improving the overall resource efficiency and economic viability of using agricultural biomass as a feedstock. Despite the importance of Si and P, the detailed ash transformation processes during entrained flow conversion of such biomass assortments remain inadequately understood. This knowledge is crucial for reducing or eliminating ash-related issues while unlocking pathways for recovering valuable Si- and P-containing compounds during entrained flow conversion of agricultural biomass. 

The main objectives of this work were, therefore, to 1) determine the ash transformation pathways of Si during entrained flow combustion of different types of Si- and P-rich agricultural biomass, 2) determine the ash transformation pathways of P during entrained flow combustion of different types of Si- and P-rich agricultural biomass, and 3) investigate the potential of extracting valuable Si- and P- containing compounds from the gas phase and/or residual ashes formed during entrained flow conversion of different agricultural biomass. 

The study combines lab-scale experiments in a laminar drop tube furnace (DTF) at 1200 °C and 1450 °C in combustion conditions (using air) and in pyrolysis conditions (using N2), with pilot-scale combustion experiments in a 150-kW powder burner connected to a horizontal ceramic-lined furnace. Three agricultural biomass types were selected to represent a range of Si and P concentrations in the selected fuels: rice husks representing Si-rich husks from certain cereal crops like rice and oat (i.e., Si-rich fuel with minor amounts of K, Ca, Mg, and P), grass representing Si- and K-rich herbaceous energy crops from grasses and residues from certain agricultural crops such as wheat straw and other cereal straws (i.e., K-Si-rich fuels with moderate amounts of Ca, Mg, and P), and brewer’s spent grains (BSG) representing P-rich grain- and seed-based agricultural biomass (i.e., P-rich fuels with a relatively high share of Si with moderate to minor Ca, Mg, and K content). All three fuels were investigated in the lab-scale DTF, whereas rice husks and BSG were examined in a 150-kW powder burner. The produced residual materials, i.e., coarse ash fractions (> 1 µm), fine particle fractions (i.e., PM1, <1 µm), chars, and deposits were morphologically and chemically characterized using SEM-EDS, XRD, ICP-AES, IC, and CHN-analysis. Thermodynamic equilibrium calculations (TECs) were employed to interpret experimental findings and theoretically assess ash transformation pathways.

Across all investigated fuels and combustion scales, Si was predominantly retained in the coarse ash fractions (>1 µm), indicating limited volatilization under the studied conditions. During the combustion of rice husks under both scales, Si present in the outer surface of the fuel formed skeleton-like coarse ash particles. Meanwhile, the Si present in the inner part of the fuel interacted with minor ash-forming elements (i.e., K, Ca, Mg, and P) and formed Si-rich molten spheres. Overall, the resulting coarse ash fractions were comprised of amorphous non-molten Si-rich particles, Si-rich melt with moderate to minor amounts of K, Ca, Mg, and P, and crystalline SiO2 (cristobalite). For grass, the results from the combustion experiments conducted under DTF conditions showed that the fuel inherent Si initially reacted with K to form molten K-silicates. The subsequent incorporation of Ca, Mg, and P into molten K-silicates led to the formation of K-Ca-Mg-rich phosphosilicate melt in the residual coarse ash fractions. Si was found in the residual coarse ash fractions mainly as amorphous K-Ca-Mg-rich phosphosilicate melt and crystalline SiO2 (quartz), Ca2MgSi2O7, CaSiO4, KCaSi3O9, Ca7(SiO4)2(PO4)2, and Ca5(SiO4)(PO4)2. For the BSG, the experiments conducted at both scales showed that the fuel inherent Si initially interacted with partially molten Ca-Mg-phosphates, and formed Ca-Mg-rich phosphosilicate melt. Si in the residual coarse ash fractions was identified as amorphous Ca-Mg-phosphosilicate and crystalline SiO2 (i.e., quartz and/or cristobalite). 

For all investigated fuels and conditions, P was primarily retained in the coarse ash fractions (> 1 µm) mainly in the form of orthophosphate compounds. A minor to moderate amounts of fuel inherent P identified in the fine particle (i.e., PM1, <1 µm) ash fractions, indicating partial volatilization of P during the investigated conditions. During combustion of rice husks across both scales, P was primarily retained in the residual coarse ash fractions and incorporated into Si-rich molten spheres with moderate to minor amounts of K, Ca, Mg, and P. Additionally, a moderate (≈ 20%)  to high (≈ 40%) share of P was detected in the PM1 fractions under studied combustion conditions. For grass fuel, DTF experiments showed that fuel-inherent P, together with Ca and Mg, interacted with molten K-silicates and formed K-Ca-Mg-rich phosphosilicate melt. P in the residual coarse ash fractions was found as K-Ca-Mg-rich phosphosilicate melt and crystalline Ca7(SiO4)2(PO4)2, Ca5(SiO4)(PO4)2, Ca5(PO4)3(OH), Ca2.89Mg0.1(PO4)2, and Ca9MgKPO4.  Furthermore, a minor (≈ 5%) to moderate (≈ 35%) amount of P was identified in the PM1 fractions at 1200 °C and 1450 °C, respectively. For BSG, the fuel inherent P (i.e., phytates) decomposed to partially molten Ca-Mg-phosphates, which subsequently interacted with Si-rich particles, leading to the formation of a Ca-Mg-rich phosphosilicate melt. In both DTF and powder burner experiments, P in the residual coarse ash fractions was primarily retained as amorphous Ca-Mg-phosphosilicate melt and crystalline Ca3Mg3(PO4)4. A minor (≈8%) to moderate (≈23%) share of P was also detected in the PM1 fractions under the investigated combustion conditions.

The combined results from TECs and experimental studies demonstrated the fuel-specific potential for recovering valuable Si- and P-containing compounds from gas and/or residual coarse ashes (>1 µm) during entrained flow conversion of different types of agricultural biomasses. For rice husks, TECs indicated that extracting valuable Si-containing compounds (e.g., SiC (s)) from the gas phase would require very high temperatures inside the flame (i.e., around 2000 °C) to volatilize a moderate amount of fuel inherent Si. Furthermore, it would require an inert cooling atmosphere and elevated surface temperatures around 1500 °C to form potentially valuable Si-containing compounds, which is challenging to achieve in practice. However, both lab- and pilot-scale combustion experiments showed the potential to extract relatively pure silica from the residual coarse ash fractions collected after entrained flow combustion of rice husks. In the case of grass, TECs did not indicate the possibility of forming valuable Si- and/or P-containing compounds from the gas phase. Regarding BSG fuel, TECs suggest that the surplus of P to Si and cations in the fuel can facilitate formation of valuable H3PO4 from the gas phase at lower surface temperatures (i.e.,< 400 °C). Moreover, the coarse ash fractions obtained during the combustion experiments of grass and BSG primarily contained different phosphosilicate melts. Further assessment is required to determine the plant availability of P in such melts.

Clinical education is the part of the nursing education wherein students gain experiential and practical knowledge in real clinical settings. It is essential because it allows nursing students to develop clinical competence by applying and integrating theoretical knowledge through patient encounters. The nursing process is a component of clinical competence and a systematic approach to providing care which involves critical thinking. The overall aim of this thesis was to explore how the nursing process can support nursing students’ learning and development of clinical competence during clinical education. The thesis takes a qualitative research approach and consists of four interrelated studies, each designed to correspond to different aspects of the overall aim: qualitative descriptive design (I), focus group methodology (II), grounded theory (III) and qualitative longitudinal intervention study (IV). Data were collected through individual interviews (I, III, IV) and focus group discussions (II, III). The included participants were nursing students in their final year (I, III, IV), supervisors in clinical settings (II, III), clinical teacher (III) and teachers in nursing education from higher education (III). Data were analysed using qualitative content analysis (I), focus group methodology (II), constant comparative analysis (III) and pattern-oriented analysis (IV). For addressing the overall aim, an interpretative synthesis was conducted to form a comprehensive understanding of the findings from all the sub-studies. 

The overall findings show that the nursing process can provide a core structure for nursing students’ learning and development of clinical competence during their clinical education. The nursing process can serve as a framework for reflection and help students give meaning to theory in practice. Through reflection and a deeper understanding of the nursing process in practice, students can adopt a more structured approach to their clinical work, which strengthens their clinical competence. However, for learning based on the nursing process to be meaningful, students must receive sufficient support and experience the necessary prerequisites. A reciprocal relationship between the student and the supervisor is significant, as both have the responsibility to create conditions for learning and development. In addition, students’ understanding and holistic view of the patient were enhanced when they were given the opportunity to consider the entire nursing process. A clinical education grounded in person-centred learning and reciprocal supervisory relationship offers a conducive environment for students to grow and enhance their clinical competence in becoming a nurse. 

In conclusion, the nursing process appears to be a valuable structure in nursing students’ learning during clinical education. Nursing students need support from supervisors and teachers to make the process meaningful in practice. These findings have implications for educators in nursing programmes and supervisors in clinical practice with regard to structuring and improving clinical education. 

Efficiently coordinating multiple robots to collaboratively complete a set of tasks requires solving two tightly coupled problems: task assignment and collision-free path planning. A widely used formalism for this setting is the Multi-Agent Path Finding (MAPF) framework, which computes conflict-free paths for agents navigating from start to goal locations on a discrete graph. Variants such as Task Assignment and Path Finding (TAPF) extend this framework by jointly determining both agent–task pairings and collision-free paths. This dissertation advances the state of the art in MAPF and its extensions by proposing new algorithms that improve computational efficiency, scalability, and real-world applicability in dynamic, physically constrained environments.

The first contribution focuses on improving the computational efficiency of MAPF. We develop an accelerated Conflict-Based Search (CBS) framework operating over structural–semantic topometric maps, where abstracted regions such as intersections and corridors replace dense grid cells. By reasoning about conflicts at the level of region occupancy in continuous time, rather than at individual grid cells, this representation drastically reduces the number of potential conflicts and search nodes while still supporting fine-grained temporal resolution. Additionally, we propose a Hybrid Priority-Based Search (HPBS) algorithm that integrates Priority-Based Search (PBS) and local CBS within a single search framework, combining the efficiency of PBS with the completeness guarantees of CBS. Together, these methods achieve significant speedups while preserving optimal or bounded-suboptimal solution quality.

The second contribution addresses the Task Assignment and Path Finding (TAPF) problem, which extends MAPF by jointly deciding how tasks are distributed among agents and how agents move to execute them. We first propose two decoupled approaches, in which task allocation and path planning are solved sequentially. The Simulated Annealing with recurrent CBS (SA-reCBS) framework combines a metaheuristic assignment process with a recurrent CBS planner to efficiently handle large-scale TAPF instances. The second decoupled method, a cluster-based task assignment and control framework, employs hierarchical clustering, the Hungarian algorithm, and a traveling salesman formulation to compute balanced task allocations and near-optimal routes. In contrast, the Conflict-Based Search with Task Sequencing (CBS-TS) framework represents a coupled formulation that integrates task assignment and path planning within a single optimization loop. By combining Mixed-Integer Linear Programming (MILP) for task sequencing with CBS for conflict resolution, CBS-TS achieves globally optimal flowtime across all agents and tasks.

The final contribution bridges the gap between discrete MAPF theory and physically realizable robot motion. Three complementary methods are proposed to achieve multi-robot motion planning under dynamic and kinematic constraints. The first, Hybrid Conflict-Based Search (HCBS), extends CBS to continuous space by embedding motion primitives directly into the search, using A* for holonomic robots and Hybrid A* for non-holonomic robots with Ackermann steering, together with geometric body-level conflict detection. The second, a Nonlinear Model Predictive Control (NMPC) framework augmented with a reactive Artificial Potential Field (APF) layer, enables real-time position tracking and adaptive collision avoidance in dynamic environments, providing robustness against disturbances and moving obstacles. The third, a trajectory optimization pipeline, refines discrete paths generated by CBS-TS into smooth, dynamically feasible trajectories within safe corridors, enforcing kinematic constraints, maintaining safety margins, and guaranteeing precise task visitation.

Collectively, these contributions establish a unified progression from discrete, combinatorial coordination to continuous, dynamically feasible multi-robot motion planning. Through extensive simulations and experiments with heterogeneous robotic platforms, the proposed methods demonstrate improved scalability, optimality, and robustness, providing a comprehensive bridge between theoretical MAPF frameworks and practical, deployable multi-robot systems.

With the global increase of hydrogen applications and industries comes an increase in the demands for the hydrogen infrastructure, where reliability is essential. Compressors are an important part of the hydrogen infrastructure, with rolling element bearings being a key component. However, hydrogen is known to significantly reduce the mechanical properties of rolling element bearings, a phenomenon called hydrogen embrittlement. Studies report that bearings which are subjected to hydrogen can fail up to ten times faster compared to bearings operating in other conditions. This detrimental effect on bearing performance has led to research within the subject of hydrogen embrittlement in bearing steels. While it is well established in research that hydrogen affects bearings negatively, several research challenges remain. This includes both the fundamental level to further understand the interaction of hydrogen with the steel as well on a more applied level when it comes to the quantification of hydrogen damage, as most of the research available in hydrogen embrittlement has a qualitative character. 

In this research, rolling sliding tribotesting using a micropitting rig (MPR) as well as full bearing testing were performed to further understand the interaction between hydrogen and bearing steel. These tribotests were combined with electrochemical hydrogen charging, thermal desorption mass spectrometry (TDMS) as well as extended material analysis to connect the wear to the hydrogen concentration as well as hydrogen trapping state in the bearing steel. The hydrogen trapping state in the material is of importance, as a lower hydrogen trapping energy is often correlated with a higher potential of hydrogen embrittlement. The results have shown that the cyclic straining of bearing steel during tribotesting can lower the hydrogen trapping energy, highlighting the increased embrittling effect of hydrogen in bearing applications. Another finding was a quantitative correlation between diffusible hydrogen concentration of bearing steel and surface-initiated damage. The hydrogen concentration of bearing steel was systematically varied by different hydrogen pre-charging conditions, which was then followed by tribotesting and wear quantification. Results revealed that at a hydrogen concentration between 1.4-2 wppm, the wear quantity doubled, while the wear mechanisms remained the same. A novel in situ hydrogen charging full thrust bearing test rig, the Hydrogen Embrittlement in Rolling Element Bearings (HERo) rig, has been developed. This test rig offers several advantages compared to test setups previously described in the literature. As a result of in situ charging, the diffusible hydrogen concentration during tribotesting is not decreasing as in conventional setups where pre-charging is performed prior to tribotesting. This allows for a more precise measurement of the influence of hydrogen on bearing steel. Furthermore, the charging in the HERo rig is performed via the backside of the bearing washer, meaning that the electrolyte is never in contact with the wear track, preventing corrosion or alteration of the surface roughness. The experiments performed with the HERo test rig show a significant decrease in bearing runtime under the influence of hydrogen by a factor of 3. While tests without hydrogen charging failed due to lubricant degradation and surface-initiated wear, hydrogen charged tests failed due to sub-surface initiated macropitting with the presence of white etching cracks. When varying hydrogen pre-charging and in situ charging procedures, it was found that longer pre-charging led to earlier bearing fatigue, while the damage mechanism stayed identical for the different hydrogen concentrations. This indicates that a critical hydrogen concentration exists not only for surface-initiated wear as mentioned above, but also for the initiation of sub-surface crack networks and macropitting. Lastly, it was found that hydrogen promoted not only the formation of white etching cracks and white etching areas, but also martensitic decay and formation of dark etching areas. 

In summary, results of the project confirm the detrimental effect of hydrogen on bearing steel, and add quantitative data to the state of knowledge, making it possible to industrially use the research data for more reliable bearings in hydrogen environments. 

This PhD thesis advances robotic autonomy by developing novel path-planning and collision-avoidance solutions that enable resilient missions in complex, unstructured real-world environments. The primary contribution is D*+, a risk-aware path planner extending the D*-lite framework for ground and aerial robots. D*+ introduces a risk layer around occupied and unknown spaces, ensuring traversable paths with safety margins while operating on imperfect maps from real data. Its dynamic mapping supports adaptive replanning, enabling exploration missions in unknown environments, and is excellent for waypoint navigation. Real-world trials with a UAV and quadrupedal robot confirm its versatility across diverse scenarios.

The second contribution is the Detect Track and Avoid Architecture (DTAA), which tackles dynamic obstacles using YOLO-based detection, Kalman filter state estimation, and a nonlinear model predictive controller (NMPC) for anticipatory avoidance maneuvers. DTAA effectively handles fast-moving objects while following D*+ paths; however, it is limited by a short predictive horizon and susceptible to local minima. To overcome these weaknesses, this thesis introduces A*+T, a distributed, time-dependent multi-agent path planner. Built on an A*framework, A*+T integrates D*+ 's risk layers and DTAA's dynamic obstacle handling, adding a temporal dimension to the planning process, enabling collision checks in time and space. The temporal dimension enables distributed autonomous robots to plan collision-free paths in shared spaces based on other robots' planned paths.

Leveraging shared paths and predicted paths from DTAA, A*+T plans collision-free paths around the dynamic obstacles. Validated through simulations and real-world experiments, A*+T enhances mission readiness for multi-agent scenarios.

Beyond these, the thesis integrates these modules into complete robotic systems, enhancing mission control for large-scale applications. Demonstrations include mining inspections (visual and gas detection) and search-and-rescue missions (locating humans/objects). These original advancements offer robust, practical solutions for robotic navigation, validated through extensive real-world testing, and contribute significantly to autonomous systems in high-stakes environments.

Stormwater runoff is a major vector for the pollution transport. Monitoring its quality is necessary for informing effective management strategies. This thesis focuses on an analytical tool increasingly utilized by both field practitioners and researchers: sensor technology. The workflows surrounding on-site sensor deployment, data handling, and interpretation involve multiple often-overlooked nuances and decisions that are not yet common practice, motivating this systematic effort.

The work began with a critical literature review that situated sensor technology within the context of urban stormwater monitoring. The review showed that sensors are frequently treated as turn-key solutions, revealing a mismatch between their perceived maturity and the actual level of methodological development in the field. It further demonstrated that the limited set of water quality parameters that can be measured directly and continuously in-situ has limited standalone value, and that the primary analytical value of sensor data emerges when it is coupled with modelling approaches, which are commonly used to derive pollutant concentration time series.

Drawing on both the reviewed literature and a multi-year field monitoring campaign incorporating continuous sensor measurements alongside sample-based laboratory analyses, this work systematically investigated the problems and limitations of in-situ water quality sensors. Two principal types of adverse effects were distinguished: loss of data and the introduction of bias and uncertainty. The results show that several of the most frequently encountered problems are amenable to post-validation correction. A comparative evaluation of simple interpolation methods and machine-learning–based reconstruction techniques indicates that interpolation is generally sufficient under moderately dynamic conditions, while machine-learning approaches offer only limited advantages for highly dynamic segments. Comparison of field sensor measurements with laboratory reference analyses revealed parameter-specific responses, with strong agreement observed for electrical conductivity, minor field-induced effects for pH, and substantial, condition-dependent bias for turbidity related to seasonal processes. The literature review indicates that uncertainties associated with analytical context are seldom systematically investigated or quantitatively reported.

Finally, this work quantified how adverse effects propagated into pollutant concentration modelling by analysing the influence of data completeness and field-induced uncertainty. Both conceptual and regression-based models were evaluated, including simple statistical and machine-learning regression models. Model performance was strongly influenced by dataset completeness and diversity, with predictive accuracy deteriorating proportionally to the magnitude of uncertainty in the data. Conceptual buildup-washoff and washoff-only models showed poor performance, whereas higher regression model performance depended primarily on the choice and combination of explanatory variables.

Autonomous inspection plays a vital role in modern robotics, supporting routine site monitoring to mapping unstructured environments. To achieve this, field robots leverage onboard perception, planning and control systems to interpret their environment and execute complex tasks autonomously and in real time.% Establish the challenges faced and the gapAs robots become increasingly adopted for long-term and large-scale inspection missions, the need for a resourceful agent becomes evident. From the integration of autonomous robots with mining infrastructure to rapid deployment of the systems for enhancing situational awareness to extended interaction between humans and robot on inspection scenes, these applications underscore the importance of a scalable, interpretable and adaptable inspection autonomy. To this end, this thesis investigates the key enablers for autonomous inspection: view planning and scene representation. Specifically, it focuses on: (a) developing a reactive view-planning strategy, (b) constructing interpretable inspection scene representation.

The first contribution introduces the method of reactive view planning for mobile robots. Primarily, the First-Look inspection framework establishes the foundation for visual inspection. It utilizes instantaneous 3D point-cloud data alongside desired photogrammetric constraints to generate an inspection view pose. The planner recursively evaluates this process to provide visual coverage of the inspection target. Building on this, the First-Look planner is extended with an exploration behaviour to form the First-Look Inspect-Explore (FLIE) framework. Through this integration, the FLIE framework addresses visual inspection of distributed and unknown inspection targets in unknown environments. Finally, the thesis investigates a hierarchical inspection framework, combining global view planning over historical maps with adaptive local replanning capability for inspection under environment uncertainty. The second contribution explores the method of constructing an actionable, hierarchical scene graph representation of inspected targets, namely 3D Layered Semantic Graph (3DLSG). Specifically, the inspection scene is composed as an ordered set of nodes and edges, organized into abstract layers and structured in a hierarchical manner. The method encodes sensor, semantic and planning information from the scene as attributes of the nodes and edge set. Through this actionable and interpretable hierarchy, the method addresses semantic-geometric path planning for navigation and informed target-selection during inspection planning.  

Overall, this thesis demonstrates deployment of autonomous robots: Unmanned Aerial Vehicles (UAVs) and Quadrupedal platforms, in subterranean and open-pit mines and outdoor urban environments. In all scenarios, field robots operate autonomously, relying solely on onboard perception, planning and control to complete their inspection tasks.