Search

The ability to traverse an unknown environment is crucial for autonomous robot operations. However, due to the limited sensing capabilities and system constraints, approaching this problem with a single robot agent can be slow, costly, and unsafe. For example, in planetary exploration missions, the wear on the wheels of a rover from abrasive terrain should be minimized at all costs as reparations are infeasible. On the other hand, utilizing a scouting robot such as a micro aerial vehicle (MAV) has the potential to reduce wear and time costs and increasing safety of a follower robot. This work proposes a novel cooperative IPP framework that allows a scout (e.g., an MAV) to efficiently explore the minimum-cost-path for a follower (e.g., a rover) to reach the goal. We derive theoretic guarantees for our algorithm, and prove that the algorithm always terminates, always finds the optimal path if it exists, and terminates early when the found path is shown to be optimal or infeasible. We show in thorough experimental evaluation that the guarantees hold in practice, and that our algorithm is 22.5% quicker to find the optimal path and 15% quicker to terminate compared to existing methods., Comment: 8 pages, 9 figures, code will be available at https://github.com/ethz-asl/scouting-ipp

The cellular environment, characterized by its intricate composition and spatial organization, hosts a variety of organelles, ranging from membrane-bound ones to membraneless structures that are formed through liquid-liquid phase separation. Cells show precise control over the position of such condensates. We demonstrate that organelle movement in external concentration gradients, diffusiophoresis, is distinct from the one of colloids because fluxes can remain finite inside the liquid-phase droplets and movement of the latter arises from incompressibility. Within cellular domains diffusiophoresis naturally arises from biochemical reactions that are driven by a chemical fuel and produce waste. Simulations and analytical arguments within a minimal model of reaction-driven phase separation reveal that the directed movement stems from two contributions: Fuel and waste are refilled or extracted locally, resulting in concentration gradients, which (i) induce product fluxes via incompressibility and (ii) result in an asymmetric forward reaction in the droplet's surroundings (as well as asymmetric backward reaction inside the droplet), thereby shifting the droplet's position. We show that the former contribution dominates and sets the direction of the movement, away from or towards fuel source and waste sink, depending on the product molecules' affinity towards fuel and waste, respectively. The mechanism thus provides a simple means to organize condensates with different composition. Particle-based simulations and systems with more complex reaction cycles corroborate the robustness and universality of this mechanism.

The use of modern software-defined radio (SDR) devices enables the implementation of efficient communication systems in numerous scenarios. Such technology comes especially handy in the context of search and rescue (SAR) systems, enabling the incorporation of additional communication data transmission into the otherwise sub-optimally used SAR bands at 121.5 and 243~MHz. In this work, we propose a novel low-complexity, energy-efficient modulation scheme that allows transmission of additional data within chirped homing signals, while still meeting the standards of international SAR systems such as COSPAS-SARSAT. The proposed method modulates information onto small deviations of the chirp slope with respect to the required unmodulated chirp, which can be easily detected at the receiver side using digital signal processing.

The rising demand for high-performing batteries requires new technological concepts. To facilitate fast charge and discharge, hierarchically structured electrodes offer short diffusion paths in the active material. However, there are still gaps in understanding the influences on the cell performance of such electrodes. Here, we employed a cell model to demonstrate that the morphology of the hierarchically structured electrode determines which electrochemical processes dictate the cell performance. The potentially limiting processes include electronic conductivity within the porous secondary particles, solid diffusion within the primary particles, and ionic transport in the electrolyte surrounding the secondary particles. Our insights enable a goal-oriented tailoring of hierarchically structured electrodes for high-power applications.

In order to operate in human environments, a robot's semantic perception has to overcome open-world challenges such as novel objects and domain gaps. Autonomous deployment to such environments therefore requires robots to update their knowledge and learn without supervision. We investigate how a robot can autonomously discover novel semantic classes and improve accuracy on known classes when exploring an unknown environment. To this end, we develop a general framework for mapping and clustering that we then use to generate a self-supervised learning signal to update a semantic segmentation model. In particular, we show how clustering parameters can be optimized during deployment and that fusion of multiple observation modalities improves novel object discovery compared to prior work. Models, data, and implementations can be found at https://github.com/hermannsblum/scim, Comment: accepted at ISRR 2022

We grafted polyacrylic acid brushes containing cysteine side chains at a defined surface density on planar lipid membranes. Specular X-ray reflectivity data indicated that the addition of Cd2+ ions induces the compaction of the polymer brush layer and modulates the adhesion of lipid vesicles. The critical threshold level inducing the switch from non-wetting to partial wetting state, [Cd2+] = 0.25 mM, was determined by microinterferometry. The interactions between vesicles and brushes were evaluated by height fluctuations of the membrane in contact with brushes and the shape of vesicles near the surface. To analyze these experiments, we have studied adhesion of axially symmetric vesicles for finite-range membrane-substrate interaction and buoyancy through simulations. We found that the local transversality condition that relates the maximal curvature at the edge of the adhesion zone to the adhesion strength remains rather accurate. Thus, although it is not experimentally possible to prepare vesicles at zero buoyancy, we can use the transversality condition to estimate the adhesion strength. However, the adhesion diagram is significantly modified by a finite range of membrane-substrate interaction and buoyancy. For downward buoyancy, vesicles merely sediment onto the substrate and there is no mean-field adhesion transition. For upward buoyancy, adhered vesicles are metastable at best. Thus, a mean-field adhesion transition can only occur at vanishing buoyancy. Only for zero-range membrane-substrate interaction does a second-order adhesion transition occur at finite interaction strength. For any finite-range interaction, the transition occurs when the membrane-substrate interaction changes from repulsive to attractive. We present a adhesion diagram as a function of adhesion strength and buoyancy and compare the adhesion behavior of vesicles to the wetting behavior of droplets of liquids., Comment: 15 pages, 17 figures

This paper presents a probabilistic framework to obtain both reliable and fast uncertainty estimates for predictions with Deep Neural Networks (DNNs). Our main contribution is a practical and principled combination of DNNs with sparse Gaussian Processes (GPs). We prove theoretically that DNNs can be seen as a special case of sparse GPs, namely mixtures of GP experts (MoE-GP), and we devise a learning algorithm that brings the derived theory into practice. In experiments from two different robotic tasks -- inverse dynamics of a manipulator and object detection on a micro-aerial vehicle (MAV) -- we show the effectiveness of our approach in terms of predictive uncertainty, improved scalability, and run-time efficiency on a Jetson TX2. We thus argue that our approach can pave the way towards reliable and fast robot learning systems with uncertainty awareness., Comment: 12 pages, 6 figures and 1 table. Accepted at the 5th Conference on Robot Learning (CORL 2021), London, UK

In the future, extraterrestrial expeditions will not only be conducted by rovers but also by flying robots. The technical demonstration drone Ingenuity, that just landed on Mars, will mark the beginning of a new era of exploration unhindered by terrain traversability. Robust self-localization is crucial for that. Cameras that are lightweight, cheap and information-rich sensors are already used to estimate the ego-motion of vehicles. However, methods proven to work in man-made environments cannot simply be deployed on other planets. The highly repetitive textures present in the wastelands of Mars pose a huge challenge to descriptor matching based approaches. In this paper, we present an advanced robust monocular odometry algorithm that uses efficient optical flow tracking to obtain feature correspondences between images and a refined keyframe selection criterion. In contrast to most other approaches, our framework can also handle rotation-only motions that are particularly challenging for monocular odometry systems. Furthermore, we present a novel approach to estimate the current risk of scale drift based on a principal component analysis of the relative translation information matrix. This way we obtain an implicit measure of uncertainty. We evaluate the validity of our approach on all sequences of a challenging real-world dataset captured in a Mars-like environment and show that it outperforms state-of-the-art approaches., Comment: Accepted to IROS 2021. Updated version corresponding to IROS camera-ready. The source code is publicly available at: https://github.com/DLR-RM/granite

A hierarchical (triple scale) simulation methodology is presented for the prediction of the dynamical and rheological properties of high molecular weight entangled polymer melts. The methodology consists of atomistic, moderately coarse-grained (mCG), and highly coarse-grained slip-spring (SLSP) simulations. At the mCG level, a few chemically bonded atoms are lumped into one coarse-grained bead. At this level, the chemical identity of the atomistic system, and the interchain topological constraints (entanglements) are preserved. The mCG potentials are derived by matching local structural distributions of the mCG model to those of the atomistic model through iterative Boltzmann inversion. For matching mCG and atomistic dynamics, the mCG time is scaled by a time scaling factor, which compensates for the lower monomeric friction coefficient of the mCG model than that of the atomistic one. At the SLSP level, multiple Kuhn segments of a polymer chain are represented by one coarse-grained bead. The very soft nonbonded interactions between beads do not prevent chain crossing and, hence, can not capture entanglements. The topological constraints are represented by slip-springs. A compensating pair potential is used in the SLSP model, to keep the static macromolecular properties unaltered upon the introduction of slip-springs. The static and kinetic parameters of the SLSP model are determined based on the lower level simulation models. Particularly, matching the orientational autocorrelation of the end-to-end vector, we determine the number of slip-springs and calibrate the timescale of the SLSP model. As the test case, the hierarchical methodology is applied to $cis$-1,4-polybutadiene (cPB) at 413 K. Dynamical properties of cPB melts are calculated for a broad range of molecular weights. The calculations are compared, and found in good agreement, with experimental data from the literature.

Graphene-based nanostructured systems and van-der-Waals heterostructures comprise a material class of growing technological and scientific importance. Joining materials with vastly different properties, polymer-graphene heterosystems promise diverse applications in surface- and nanotechnology, including photovoltaics or nanotribology. Fundamentally, molecular adsorbates are prototypical systems to study confinement-induced phase transitions exhibiting intricate dynamics, which require a comprehensive understanding of the dynamical and static properties on molecular time and length scales. Here, we investigate the dynamics and the structure of a single polyethylene chain on free-standing graphene by means of molecular dynamics simulations. In equilibrium, the adsorbed polymer is orientationally linked to the graphene as two-dimensional folded-chain crystallites or, at elevated temperatures, as a floating solid. The associated superstructure can be reversibly melted on a picosecond time scale upon quasi-instantaneous substrate heating, involving ultrafast heterogeneous melting via a transient floating phase. Our findings elucidate time-resolved molecular-scale ordering and disordering phenomena in individual polymers interacting with solids, yielding complementary information to collective friction and viscosity, and linking to recent experimental observables from ultrafast electron diffraction. We anticipate that the approach will help in resolving non-equilibrium phenomena of hybrid polymeric systems over a broad range of time and length scales.

In equilibrium the interface potential that describes the interaction between two AB interfaces in a binary blend of A and B homopolymers is attractive at all distances, resulting in coarsening of the blend morphology even in the absence of interface curvature. We demonstrate that the dissipative assembly in response to a time-periodic variation of the blend incompatibility qualitatively alters this behavior, i.e., for suitable parameters the interface potential exhibits a periodic spatial modulation and AB interfaces adopt a well-defined distance. We explore for which oscillation periods and amplitudes an interface repulsion occurs and demonstrate that we can control the preferred interface distance over a wide range by varying the oscillation period. Using particle-based simulations we explicitly demonstrate that this dissipative assembly of a homopolymer blend results in a lamellar structure with multiple planar interfaces in a thin film geometry.

Multi-component polymer systems are important for the development of new materials because of their ability to phase-separate or self-assemble into nano-structures. The Single-Chain-in-Mean-Field (SCMF) algorithm in conjunction with a soft, coarse-grained polymer model is an established technique to investigate these soft-matter systems. Here we present an im- plementation of this method: SOft coarse grained Monte-carlo Accelera- tion (SOMA). It is suitable to simulate large system sizes with up to billions of particles, yet versatile enough to study properties of different kinds of molecular architectures and interactions. We achieve efficiency of the simulations commissioning accelerators like GPUs on both workstations as well as supercomputers. The implementa- tion remains flexible and maintainable because of the implementation of the scientific programming language enhanced by OpenACC pragmas for the accelerators. We present implementation details and features of the program package, investigate the scalability of our implementation SOMA, and discuss two applications, which cover system sizes that are difficult to reach with other, common particle-based simulation methods.

Carnivores are important components of ecosystems with wide-ranging effects on ecological communities.We studied the carnivore community in the Apostle Islands National Lakeshore (APIS), where the presence, distribution, and populations of carnivores was largely unknown. We developed a systematic method to deploy camera traps across a grid while targeting fine-scale features to maximize carnivore detection (Appendix 1), including systematic methods for organizing and tagging the photo data (Appendix 2). We deployed 88 cameras on 13 islands from 2014-2016. We collected 92,694 photographs across 18,721 trap nights, including 3,591 wildlife events and 1,070 carnivore events. We had a mean of 6.6 cameras per island (range 2-30), and our camera density averaged 1.23 (range 0.74-3.08) cameras/ km2. We detected 27 species and 10 terrestrial carnivores, including surprising detections of American martens (Martes americana) and gray wolves (Canis lupus). The mean richness of carnivores on an island was 3.23 (range 0-10). The best single variable to explain carnivore richness on the Apostle Islands was island size, while the best model was island size (positive correlation) and distance from mainland (negative correlation) (R2 = 0.92). Relative abundances for carnivores ranged from a low of 0.01 for weasels (Mustela spp.) to a high of 2.64 for black bears (Ursus americanus), and the relative abundance of a species was significantly correlated with the number of islands on which they were found. Carnivore occupancy ranged from lows of 0.09 for gray wolves and 0.11 for weasels to a high of 0.82 for black bears. Fuller understanding of APIS ecology will require on-going monitoring of carnivores to evaluate temporal dynamics as well as related ecological evaluations (e.g. small mammal dynamics, plant community dynamics) to understand trophic effects.