Repository for Publications and Research Data
News from the ETH Library
Recently Added
First Continuous Aerosol Measurements at Testa Grigia at 3480 m a.s.l.: Aerosol Populations and Dust Transport Dynamics in the Southern European Alps
Item type: Journal Article
Gilardoni, Stefania; Bellini, Annachiara; Bonasoni, Paolo; et al. (2026)
High-elevation observatories are crucial for monitoring atmospheric aerosols, which play a key role in the climate system due to their effects on radiation, cloud, and snow albedo. We present the first measurements of aerosol size distribution and absorption coefficient at a 1 h time resolution collected at the Testa Grigia Observatory (3480 m a.s.l.) in the Italian Alps. This dataset spans from September 2021 to May 2023. We identified three distinct aerosol population types reaching the observatory, reflecting distinct transport pathways. The coarse particle population is indicative of long-range transport of air masses from the Sahara. Conversely, the fine particle population is linked to mesoscale circulations and boundary layer dynamics (from the Po Valley and local alpine valleys), and broader continental flows. Finally, periods of generally low particle number correspond to the influence of clean air from the free troposphere and the Mediterranean basin. The upper bound of the frequency of boundary layer influence is equal to 28 %. Conversely, Sahara Dust Events (SDE), identified as periods characterized by coarse aerosol population transported from the Sahara region, are observed for 6 % of the time. These events are predominantly recorded during spring and early summer and show strong correspondence with reanalysis data provided by CAMS (Copernicus Atmosphere Monitoring Service) ensemble model. The seasonal variability of PM10 concentration associated to SDE is explained by the sensitivity to dust emission regions, dust mobilization over source region, and efficiency of dust transport mechanisms.
Femtosecond demagnetization: Towards ultrafast spintronics
Item type: Habilitation Thesis
Acremann, Yves Marc (2022)
Préparer la guerre: Militarisation de la jeunesse russe
Item type: Other Publication
Perović, Jeronim (2026)
Les jeunes sont une cible centrale de la militarisation russe. À travers les écoles, les organisations de jeunesse et les centres de formation, l’État prépare la prochaine génération à la guerre.
Vorbereitung auf den Krieg: Militarisierung der russischen Jugend
Item type: Other Publication
Perović, Jeronim (2026)
Junge Menschen sind zu einer zentralen Zielgruppe der Militarisierung Russlands geworden. Über Schulen, Jugendorganisationen und Ausbildungszentren versucht der Staat, Patriotismus zu vermitteln, Loyalität zu fördern und die nächste Generation auf Krieg vorzubereiten.
State Estimation and System Design for Precision-Critical Aerial Interaction
Item type: Doctoral Thesis
Lanegger, Christian (2026)
At the beginning of this thesis, quadcopter research was expanding rapidly, with many academic advances transferred directly into industry.
While their mechanical simplicity and limited actuation made them well suited for inspection and surveillance, a new research direction emerged to extend their capabilities through additional actuation.
These aerial robots can exert forces in arbitrary directions and, therefore, can physically interact with their environment, enabling applications such as non-destructive testing (NDT), infrastructure maintenance, and construction work at height.
However, aerial robots are significantly more complex than classical quadcopters.
Their applications demand higher precision and require deployment in often confined, cluttered, and changing environments rather than open, GNSS-supported spaces.
This increases the requirements on state estimation, which must be more precise, as well as resilient to changing environmental conditions and disturbances.
A central challenge arises from the inherent trade-off between precision and reliability: systems optimized for accuracy under laboratory conditions often fail in the presence of unforeseen disturbances, while systems prioritizing reliability frequently have to sacrifice precision.
Moreover, task requirements, such as interaction forces, allowable position errors, or reliable sensing modalities, vary substantially.
This makes state estimation and overall system performance strongly context-dependent.
Contributing to aerial robotics in a way that advances the state-of-the-art while remaining relevant for real-world deployment is therefore not straightforward.
This thesis addresses these challenges through application-driven research.
By developing an aerial robotic platform for layouting on construction site ceilings, we identified key obstacles preventing reliable real-world deployment and addressed them in four contributions.
First, we demonstrate that through dedicated hardware design, aerial robots possess the capabilities to meet industrial precision requirements.
In an ablation study on a floating-base aerial manipulator, we identified three key features for precise physical interaction: multiple contact points, active end-effector actuation, and sufficient compliance to generate internal forces and compensate for pose errors and disturbances.
Integrating these features enabled millimeter-level accuracy (1.5–3.5mm mean absolute error) under laboratory conditions and accurate ground-truth pose information.
In a second step, we deployed the system on a mock-up construction site, which exposed the trade-off between reliability and accuracy in state estimation.
When tightly fusing accurate but dropout-prone total-station measurements with reliable but less accurate visual-inertial odometry (VIO) estimates, the system could not guarantee millimeter precision and reliable operation.
We resolve this conflict through a polylithic state-estimation architecture that separates objectives: a local estimator ensures stable flight by providing high-rate, smooth estimates, while a high-accuracy global estimator focuses on drift correction to enable precise task execution.
By explicitly separating reliability and precision at the architectural level, the system can switch between operating modes depending on task requirements and estimate quality.
This resolves the apparent trade-off not by compromise, but by structural design.
However, switching between operating modes autonomously requires reliable knowledge of the quality of the underlying state estimates.
Without this awareness, the system cannot decide when precision is sufficient for task execution or when safe flight must take precedence.
Common quality indicators such as covariance estimates often fail to reflect true uncertainty, as they rely on simplified or inaccurate sensor models that do not hold in real-world conditions.
In the final two contributions of this thesis, we therefore introduce two novel methods for estimator introspection that quantify the reliability of state estimates in a sensor-agnostic manner.
The first detects disagreements by comparing the distributions of recent velocity estimates from independent estimators using the Cramér von Mises (CM) distance.
While effective, maintaining multiple parallel estimators increases system complexity and is particularly challenging on aerial robots with limited payload and sensing resources.
To address this limitation, we further propose an introspective method that evaluates the trustworthiness of an inertial measurement unit (IMU)-based sensor fusion framework directly through spectral analysis.
We demonstrate that degradation in inertial sensing can be identified from the spectral characteristics of the estimated velocity signal, providing a self-contained indicator of estimator reliability without external reference.
To sum up, this thesis demonstrates that enabling aerial robots to deliver real-world value requires co-design of hardware, estimation, and system architecture.
By starting from a concrete application, previously hidden system-level challenges become visible, leading to solutions that advance both scientific understanding and practical usability.
At the same time, the rapid industrial and recent military adoption of drone technology highlights the dual-use nature of aerial robotics.
In application-driven research, this dual-use character places additional responsibility on researchers to remain aware of the societal implications of their work and to critically reflect on which applications they choose to enable and prioritize.