ETH Zurich

With the introduction of automated vehicles, new operating regimes for public transport services will become possible. A station-based Automated Transit on Demand service could be an attractive alternative to the current modes of transportation. In this paper, the impact of this kind of service on the modal share for the city of Zurich, Switzerland, and its surrounding area is modeled using an agent-based approach. Different scenarios regarding the operating area, pricing scheme, and a cordon charge are tested on their potential to make use of the benefits of the new service while preventing an overflow of automated vehicles in the urban core. Results show that if left unconstrained the proposed service can substantially impact the demand for public transport. A pricing scheme that bases the pricing of the new service relative to the accessibility of the current public transport service is a promising solution to increase the accessibility of the rural areas while maintaining a high modal share for public transport in the city center. Finally, using an optimization algorithm we show that the total car-fleet and public parking space can be reduced at the cost of a slight increase in vehicle kilometers traveled. Moreover, we find that the cost coverage of the proposed transit service is potentially much higher in comparison to current public transport services.

Robotic exploration or monitoring missions require mobile robots to autonomously and safely navigate between multiple target locations in potentially challenging environments. Currently, this type of multi-goal mission often relies on humans designing a set of actions for the robot to follow in the form of a path or waypoints. In this work, we consider the multi-goal problem of visiting a set of pre-defined targets, each of which could be visited from multiple potential locations. To increase autonomy in these missions, we propose a safe multi-goal (SMUG) planner that generates an optimal motion path to visit those targets. To increase safety and efficiency, we propose a hierarchical state validity checking scheme, which leverages robot-specific traversability learned in simulation. We use LazyPRM* with an informed sampler to accelerate collision-free path generation. Our iterative dynamic programming algorithm enables the planner to generate a path visiting more than ten targets within seconds. Moreover, the proposed hierarchical state validity checking scheme reduces the planning time by 30% compared to pure volumetric collision checking and increases safety by avoiding high-risk regions. We deploy the SMUG planner on the quadruped robot ANYmal and show its capability to guide the robot in multi-goal missions fully autonomously on rough terrain.

In this paper, we consider the problem of system identification when side-information is available on the steady-state gain (SSG) of the system. We formulate a general nonparametric identification method as an infinite-dimensional constrained convex program over the reproducing kernel Hilbert space (RKHS) of stable impulse responses. The objective function of this optimization problem is the empirical loss regularized with the norm of RKHS, and the constraint is considered for enforcing the integration of the SSG side-information. The proposed formulation addresses both the discrete-time and continuous-time cases. We show that this program has a unique solution obtained by solving an equivalent finite-dimensional convex optimization. This solution has a closed-form when the empirical loss and regularization functions are quadratic and exact side-information is considered. We perform extensive numerical comparisons to verify the efficiency of the proposed identification methodology.

Run-time resource management is fundamental for efficient execution of workloads on Chip Multiprocessors. Application- and system-level requirements (e.g. on performance vs. power vs. lifetime reliability) are generally conflicting each other, and any decision on resource assignment, such as core allocation or frequency tuning, may positively affect some of them while penalizing some others. Resource assignment decisions can be perceived in few instants of time on performance and power consumption, but not on lifetime reliability. In fact, this latter changes very slowly based on the accumulation of effects of various decisions over a long time horizon. Moreover, aging mechanisms are various and have different causes; most of them, such as Electromigration (EM), are subject to temperature levels, while Thermal Cycling (TC) is caused mainly by temperature variations (both amplitude and frequency). Mitigating only EM may negatively affect TC and vice versa. We propose a resource orchestration strategy to balance the performance and power consumption constraints in the short-term and EM and TC aging in the long-term. Experimental results show that the proposed approach improves the average Mean Time To Failure at least by 17% and 20% w.r.t. EM and TC, respectively, while providing same performance level of the nominal counterpart and guaranteeing the power budget.

A highly energy-efficient neuromorphic computing-in-memory (Neuro-CIM) processor is proposed for ultralow-power deep learning applications. Neuro-CIM can support spiking neural network (SNN) to eliminate the power and area overhead of previous CIM processor. The sign extended bits gating reduces the bitline (BL) voltage switching rate due to negative small-magnitude weights allowing 38% power reduction at 8-b weight condition and 25% at 4-b weight condition. In addition, Neuro-CIM replaces high-precision analog-to-digital converter (ADC) with 1-b comparator by exploiting the characteristic of the SNN, and thus, power and area efficiencies are significantly enhanced. Furthermore, the early stopping scheme terminates unnecessary neuronal operations, reducing power consumption by 31%. In addition, the analog and digital networks are integrated for high reconfigurability and energy efficiency. The analog network with voltage folding circuit enables accurate analog-domain aggregation by increasing the dynamic range without compromising the voltage resolution. The digital network-in-memory supports input–output channel extension for high reconfigurability and input data reuse scheme for reducing input memory (IMEM) access. Neuro-CIM is fabricated in 28-nm CMOS technology and occupies the 2.9-mm $^{2}$ die area. It achieves the state-of-the-art energy consumption per classification of 0.72 $\mu$ J and 92.1% accuracy for CIFAR-10 with 4-b input and 4-b weight and 372.2 $\mu$ J and 65.8% accuracy for ImageNet with 6-b input and 8-b weight at 200 MHz, and 1.1-V conditions. Moreover, the proposed CIM processor achieves 310.37 tera operations per second/watt (TOPS/W) and 90.7% accuracy with 4-b input and 1-b weight for Canadian Institute for Advanced Research, 10 classes (CIFAR-10) classification.

This brief proposes a 250 GHz wideband mixer-first and direct-conversion RX adopting a baseband (BB) equalized single-balanced resistive mixer. The conventional resistive mixer suffers from the trade-off between gain and bandwidth, which hinders the wideband and low-loss mixer implementation. To address this challenge, the proposed RX employs the BB equalization technique that compensates for the narrowband response of the low-loss mixer by cascading it with the deliberately shaped response of the following BB amplifier, thereby achieving low-loss yet wideband characteristics of the overall receiver chain. Moreover, the conversion gain and the noise figure of the single-balanced mixer are further improved by adopting a λ/4 open stub that suppresses the undesired LO leakage. Implemented in a 65 nm CMOS, the proposed RX achieves an effective 3-dB bandwidth of 26 GHz while dissipating a total dc power of 132 mW. Based on the calculated signal to noise-and-distortion ratio (SNDR), the proposed RX is expected to support a 16QAM demodulation with a data rate higher than 100 Gbps, in principle.

Real-time detection of moving objects is an essential capability for robots acting autonomously in dynamic environments. We thus propose Dynablox , a novel online mapping-based approach for robust moving object detection in complex unstructured environments. The central idea of our approach is to incrementally estimate high confidence free-space areas by modeling and accounting for sensing, state estimation, and mapping limitations during online robot operation. The spatio-temporally conservative free space estimate enables robust detection of moving objects without making any assumptions on the appearance of objects or environments. This allows deployment in complex scenes such as multi-storied buildings or staircases, and for diverse moving objects such as people carrying various items, doors swinging or even balls rolling around. We thoroughly evaluate our approach on real-world data sets, achieving 86% IoU at 17 FPS in typical robotic settings. The method outperforms a recent appearance-based classifier and approaches the performance of offline methods. We demonstrate its generality on a novel data set with rare moving objects in complex environments. We make our efficient implementation and the novel data set available as open-source.

The shuttle pump is a novel implantable total artificial heart (TAH) concept based on a Linear-Rotary Actuator (LiRA) and currently under development at the Power Electronic Systems Laboratory, ETH Zurich in close partnership with Charité Berlin and the Medical University of Vienna. This paper presents the analysis, design and realization of the ShuttlePump Linear Actuator (LA) part, which is necessary to provide about 45 N of axial actuation force. Design criteria are minimization of volume and generated power losses in the winding, which could result in excess heating and/or blood damage, i.e. protein denaturation and aggregation. The LA is implemented as a Tubular LA (TLA) to maximize the active area for linear/axial force generation. After a preliminary analysis based on first principles, the TLA is optimized in detail with the aid of FEM simulations. The experimental measurements conducted on the realized TLA prototype verify the FEM simulation results and confirm the suitability for the realization of the ShuttlePump TAH.

Robot-assisted neurorehabilitation requires trajectories between arbitrary poses in the patient's range of motion. Data-driven optimization methods, such as Learning by Demonstration, are well suited to replicate complex multi-joint movements. However, these methods lack individualization to patient-, robot- and exercise-specific constraints. We propose a hybrid optimization framework that combines cost-based objectives, such as minimizing jerk, with the data-driven optimization of a reference trajectory. The objectives can be individually weighted in a sequential quadratic program with application-related constraints represented in intuitive workspaces. We demonstrated that trajectories recorded from an existing upper-limb activity dataset could be adapted to the personal needs of a healthy participant with simulated impairments, the hardware-specific robot topology, and changes in the exercise setup. Furthermore, we showed how redundancies in the degrees of freedom of the arm can be exploited: For example, an elbow angle movement of $30.4^{\circ }$ was compensated entirely through increased wrist movement in a reach-goal task. In addition to making sequential quadratic programming more accessible to the field of rehabilitation robotics, our framework improves the variability and individualizability of generated trajectories for patients, provides more adaptation possibilities to the therapist, and enables sharing of recorded movement data between robotic platforms, patients, and exercises.

Targeted eradication of transformed or otherwise dysregulated cells using monoclonal antibodies (mAb), antibody–drug conjugates (ADC), T cell engagers (TCE), or chimeric antigen receptor (CAR) cells is very effective for hematologic diseases. Unlike the breakthrough progress achieved for B cell malignancies, there is a pressing need to find suitable antigens for myeloid malignancies. CD123, the interleukin-3 (IL-3) receptor alpha-chain, is highly expressed in various hematological malignancies, including acute myeloid leukemia (AML). However, shared CD123 expression on healthy hematopoietic stem and progenitor cells (HSPCs) bears the risk for myelotoxicity. We demonstrate that epitope-engineered HSPCs were shielded from CD123-targeted immunotherapy but remained functional, while CD123-deficient HSPCs displayed a competitive disadvantage. Transplantation of genome-edited HSPCs could enable tumor-selective targeted immunotherapy while rebuilding a fully functional hematopoietic system. We envision that this approach is broadly applicable to other targets and cells, could render hitherto undruggable targets accessible to immunotherapy, and will allow continued posttransplant therapy, for instance, to treat minimal residual disease (MRD).

Marine sediments comprise the primary long‐term sink of organic matter (OM) in marine systems. Disentangling the diverse origins of OM and the influence of the main processes that determine organic carbon (OC) fate at a global scale has proven difficult due to limited spatial data coverage. Thus, comprehensive studies of the spatial distribution of the content and geochemical characteristics of sedimentary OM at basin scales provide fundamental knowledge on the role of marine sediments in the global carbon cycle. Here, we shed light on the origin of OM and the underlying mechanisms that determine its fate in a semi‐enclosed basin by examining the spatial patterns in the isotopic and elemental composition of OM in 149 core‐top samples from the Western Mediterranean Sea and the adjacent Atlantic Ocean sector. Our results reveal an apparent SW‐NE gradient that reverses in the Gulf of Lions in most geochemical and sedimentological features. Changes in the OC content and ẟ ¹³ C and Δ ¹⁴ C signatures are ascribed to spatial variations in marine primary productivity and the influence of varying discharge of rivers and well‐developed canyons that favour the cross‐shelf transport of terrestrial (and petrogenic) OC. Our results also suggest the potential influence of two other mechanisms on the geochemical signatures of OM: i) lateral transport of allochthonous OC and selective degradation of labile OM, which potentially occurs across the studied area having a greater impact towards the north‐eastern region, and ii) OM protection via association with mineral surfaces, potentially having a greater influence towards the south‐western basins. This article is protected by copyright. All rights reserved.

In this paper we review the academic transportation literature published between 2014 and 2018 to evaluate where the field stands regarding the use and misuse of statistical significance in empirical analysis, with a focus on discrete choice models. Our results show that 39% of studies explained model results exclusively based on the sign of the coefficient, 67% of studies did not distinguish statistical significance from economic, policy or scientific significance in their conclusions, and none of the reviewed studies considered the statistical power of the tests. Based on these results we put forth a set of recommendations aimed at shifting the focus away from statistical significance towards proper and comprehensive assessment of effect magnitudes and other policy relevant quantities.

We prove that the mesoscopic linear statistics \(\sum _i f(n^a(\sigma _i-z_0))\) of the eigenvalues \(\{\sigma _i\}_i\) of large \(n\times n\) non-Hermitian random matrices with complex centred i.i.d. entries are asymptotically Gaussian for any \(H^{2}_0\)-functions \(f\) around any point \(z_0\) in the bulk of the spectrum on any mesoscopic scale \(0<a<1/2\). This extends our previous result (Cipolloni et al. in Commun Pure Appl Math, 2019. arXiv:1912.04100), that was valid on the macroscopic scale, \(a=0\), to cover the entire mesoscopic regime. The main novelty is a local law for the product of resolvents for the Hermitization of X at spectral parameters \(z_1, z_2\) with an improved error term in the entire mesoscopic regime \(|z_1-z_2|\gg n^{-1/2}\). The proof is dynamical; it relies on a recursive tandem of the characteristic flow method and the Green function comparison idea combined with a separation of the unstable mode of the underlying stability operator.

The worldwide variation in vegetation height is fundamental to the global carbon cycle and central to the functioning of ecosystems and their biodiversity. Geospatially explicit and, ideally, highly resolved information is required to manage terrestrial ecosystems, mitigate climate change and prevent biodiversity loss. Here we present a comprehensive global canopy height map at 10 m ground sampling distance for the year 2020. We have developed a probabilistic deep learning model that fuses sparse height data from the Global Ecosystem Dynamics Investigation (GEDI) space-borne LiDAR mission with dense optical satellite images from Sentinel-2. This model retrieves canopy-top height from Sentinel-2 images anywhere on Earth and quantifies the uncertainty in these estimates. Our approach improves the retrieval of tall canopies with typically high carbon stocks. According to our map, only 5% of the global landmass is covered by trees taller than 30 m. Further, we find that only 34% of these tall canopies are located within protected areas. Thus, the approach can serve ongoing efforts in forest conservation and has the potential to foster advances in climate, carbon and biodiversity modelling.

High-harmonic spectroscopy is an all-optical nonlinear technique with inherent attosecond temporal resolution. It has been applied to a variety of systems in the gas phase and solid state. Here we extend its use to liquid samples. By studying high-harmonic generation over a broad range of wavelengths and intensities, we show that the cut-off energy is independent of the wavelength beyond a threshold intensity and that it is a characteristic property of the studied liquid. We explain these observations with a semi-classical model based on electron trajectories that are limited by the electron scattering. This is further confirmed by measurements performed with elliptically polarized light and with ab-initio time-dependent density functional theory calculations. Our results propose high-harmonic spectroscopy as an all-optical approach for determining the effective mean free paths of slow electrons in liquids. This regime is extremely difficult to access with other methodologies, but is critical for understanding radiation damage to living tissues. Our work also indicates the possibility of resolving subfemtosecond electron dynamics in liquids offering an all-optical approach to attosecond spectroscopy of chemical processes in their native liquid environment.

Understanding and predicting molecular responses in single cells upon chemical, genetic or mechanical perturbations is a core question in biology. Obtaining single-cell measurements typically requires the cells to be destroyed. This makes learning heterogeneous perturbation responses challenging as we only observe unpaired distributions of perturbed or non-perturbed cells. Here we leverage the theory of optimal transport and the recent advent of input convex neural architectures to present CellOT, a framework for learning the response of individual cells to a given perturbation by mapping these unpaired distributions. CellOT outperforms current methods at predicting single-cell drug responses, as profiled by scRNA-seq and a multiplexed protein-imaging technology. Further, we illustrate that CellOT generalizes well on unseen settings by (1) predicting the scRNA-seq responses of holdout patients with lupus exposed to interferon-β and patients with glioblastoma to panobinostat; (2) inferring lipopolysaccharide responses across different species; and (3) modeling the hematopoietic developmental trajectories of different subpopulations.

Collectives form nonequilibrium social structures characterized by volatile dynamics. Individuals join or leave. Social relations change quickly. Therefore, unlike engineered or ecological systems, a resilient reference state cannot be defined. We propose a novel resilience measure combining two dimensions: robustness and adaptivity. We demonstrate how they can be quantified using data from a software-developer collective. Our analysis reveals a resilience life cycle (i.e., stages of increasing resilience are followed by stages of decreasing resilience). We explain the reasons for these observed dynamics and provide a formal model to reproduce them. The resilience life cycle allows distinguishing between short-term resilience, given by a sequence of resilient states, and long-term resilience, which requires collectives to survive through different cycles.

First-passage time statistics in disordered systems exhibiting scale invariance are studied widely. In particular, long trapping times in energy or entropic traps are fat-tailed distributed, which slow the overall transport process. We study the statistical properties of the first-passage time of biased processes in different models, and we employ the big-jump principle that shows the dominance of the maximum trapping time on the first-passage time. We demonstrate that the removal of this maximum significantly expedites transport. As the disorder increases, the system enters a phase where the removal shows a dramatic effect. Our results show how we may speed up transport in strongly disordered systems exploiting scale invariance. In contrast to the disordered systems studied here, the removal principle has essentially no effect in homogeneous systems; this indicates that improving the conductance of a poorly conducting system is, theoretically, relatively easy as compared to a homogeneous system.