Search

Embodied navigation requires robots to understand and interact with the environment based on given tasks. Vision-Language Navigation (VLN) is an embodied navigation task, where a robot navigates within a previously seen and unseen environment, based on linguistic instruction and visual inputs. VLN agents need access to both local and global action spaces; former for immediate decision making and the latter for recovering from navigational mistakes. Prior VLN agents rely only on instruction-viewpoint alignment for local and global decision making and back-track to a previously visited viewpoint, if the instruction and its current viewpoint mismatches. These methods are prone to mistakes, due to the complexity of the instruction and partial observability of the environment. We posit that, back-tracking is sub-optimal and agent that is aware of its mistakes can recover efficiently. For optimal recovery, exploration should be extended to unexplored viewpoints (or frontiers). The optimal frontier is a recently observed but unexplored viewpoint that aligns with the instruction and is novel. We introduce a memory-based and mistake-aware path planning strategy for VLN agents, called \textit{StratXplore}, that presents global and local action planning to select the optimal frontier for path correction. The proposed method collects all past actions and viewpoint features during navigation and then selects the optimal frontier suitable for recovery. Experimental results show this simple yet effective strategy improves the success rate on two VLN datasets with different task complexities.

Embodied AI aims to develop robots that can \textit{understand} and execute human language instructions, as well as communicate in natural languages. On this front, we study the task of generating highly detailed navigational instructions for the embodied robots to follow. Although recent studies have demonstrated significant leaps in the generation of step-by-step instructions from sequences of images, the generated instructions lack variety in terms of their referral to objects and landmarks. Existing speaker models learn strategies to evade the evaluation metrics and obtain higher scores even for low-quality sentences. In this work, we propose SAS (Spatially-Aware Speaker), an instruction generator or \textit{Speaker} model that utilises both structural and semantic knowledge of the environment to produce richer instructions. For training, we employ a reward learning method in an adversarial setting to avoid systematic bias introduced by language evaluation metrics. Empirically, our method outperforms existing instruction generation models, evaluated using standard metrics. Our code is available at \url{https://github.com/gmuraleekrishna/SAS}.

Design of an efficient thread-safe concurrent data structure is a balancing act between its implementation complexity and performance. Lock-based concurrent data structures, which are relatively easy to derive from their sequential counterparts and to prove thread-safe, suffer from poor throughput under even light multi-threaded workload. At the same time, lock-free concurrent structures allow for high throughput, but are notoriously difficult to get right and require careful reasoning to formally establish their correctness. We explore a solution to this conundrum based on batch parallelism, an approach for designing concurrent data structures via a simple insight: efficiently processing a batch of a priori known operations in parallel is easier than optimising performance for a stream of arbitrary asynchronous requests. Alas, batch-parallel structures have not seen wide practical adoption due to (i) the inconvenience of having to structure multi-threaded programs to explicitly group operations and (ii) the lack of a systematic methodology to implement batch-parallel structures as simply as lock-based ones. We present OBatcher-an OCaml library that streamlines the design, implementation, and usage of batch-parallel structures. It solves the first challenge (how to use) by suggesting a new lightweight implicit batching design that is built on top of generic asynchronous programming mechanisms. The second challenge (how to implement) is addressed by identifying a family of strategies for converting common sequential structures into efficient batch-parallel ones. We showcase OBatcher with a diverse set of benchmarks. Our evaluation of all the implementations on large asynchronous workloads shows that (a) they consistently outperform the corresponding coarse-grained lock-based implementations and that (b) their throughput scales reasonably with the number of processors., Comment: Extended version of the OOPSLA'24 paper

At first light, the Thirty Meter Telescope (TMT) near-infrared (NIR) instruments will be fed by a multiconjugate adaptive optics instrument known as the Narrow Field Infrared Adaptive Optics System (NFIRAOS). NFIRAOS will use six laser guide stars to sense atmospheric turbulence in a volume corresponding to a field of view of 2', but natural guide stars (NGSs) will be required to sense tip/tilt and focus. To achieve high sky coverage (50% at the north Galactic pole), the NFIRAOS client instruments use NIR on-instrument wavefront sensors that take advantage of the sharpening of the stars by NFIRAOS. A catalog of guide stars with NIR magnitudes as faint as 22 mag in the J band (Vega system), covering the TMT-observable sky, will be a critical resource for the efficient operation of NFIRAOS, and no such catalog currently exists. Hence, it is essential to develop such a catalog by computing the expected NIR magnitudes of stellar sources identified in deep optical sky surveys using their optical magnitudes. This paper discusses the generation of a partial NIR Guide Star Catalog (IRGSC), similar to the final IRGSC for TMT operations. The partial catalog is generated by applying stellar atmospheric models to the optical data of stellar sources from the Panoramic Survey Telescope and Rapid Response System (Pan-STARRS) optical data and then computing their expected NIR magnitudes. We validated the computed NIR magnitudes of the sources in some fields by using the available NIR data for those fields. We identified the remaining challenges of this approach. We outlined the path for producing the final IRGSC using the Pan-STARRS data. We have named the Python code to generate the IRGSC as irgsctool, which generates a list of NGS for a field using optical data from the Pan-STARRS 3pi survey and also a list of NGSs having observed NIR data from the UKIRT Infrared Deep Sky Survey if they are available.

We present new near-ultraviolet (NUV, $\lambda$ = 2479 $-$ 3306 $\r{A}$) transmission spectroscopy of KELT-9b, the hottest known exoplanet, obtained with the Colorado Ultraviolet Transit Experiment ($CUTE$) CubeSat. Two transits were observed on September 28th and September 29th 2022, referred to as Visits 1 and 2 respectively. Using a combined transit and systematics model for each visit, the best-fit broadband NUV light curves are R$_{\text{p}}$/R$_{\star}$ $=$ 0.136$_{0.0146}^{0.0125}$ for Visit 1 and R$_{\text{p}}$/R$_{\star}$ $=$ 0.111$_{0.0190}^{0.0162}$ for Visit 2, appearing an average of 1.54$\times$ larger in the NUV than at optical wavelengths. While the systematics between the two visits vary considerably, the two broadband NUV light curves are consistent with each other. A transmission spectrum with 25 $\r{A}$ bins suggests a general trend of excess absorption in the NUV, consistent with expectations for ultra-hot Jupiters. Although we see an extended atmosphere in the NUV, the reduced data lack the sensitivity to probe individual spectral lines.

Dust polarisation observations at optical wavelengths help understand the dust grain properties and trace the plane-of-the-sky component of the magnetic field. In this study, we make use of published optical polarisation data acquired with AIMPOL along with distances ($d$) and extinction ($A_{\mathrm{V}}$) data. We study the variation of polarisation efficiency ($P/A_{\mathrm{V}}$) as a function of $A_{\mathrm{V}}$ in the diffuse interstellar medium (ISM) and intracluster mediums (ICM) using the already published polarisation data of six clusters. Among these clusters, NGC 2281, NGC 1664, and NGC 1960 are old; while Stock 8, NGC 1931, and NGC 1893 are young. We categorize stars towards each cluster into foreground, background, and cluster members by employing two clustering algorithms GMM and DBSCAN. Thus, classified field stars and cluster members are used to reveal the polarisation properties of ISM and ICM dust, respectively. We find that the dust grains located in the diffuse ISM show higher polarisation efficiencies when compared to those located in the ICM of younger clusters., Comment: 12 pages, 10 figures

We present a detailed structural and gas kinematic study of the star-forming complex W5-NW. A cloud-cloud collision scenario unravels with evidences of collision induced star and cluster formation. Various signatures of cloud-cloud collision such as "complementary distribution" and "bridging-features" are explored. At the colliding region, the two clouds have complementary morphologies, where W5-NWb has a filamentary key-like shape which fits into the U-shaped cavity in W5-NWa that behaves like a keyhole. The interaction region between the two clouds is characterised by bridging features with intermediate velocities connecting the two clouds. A skewed V-shaped bridging feature is also detected at the site of collision. A robust picture of the molecular gas distribution highlighting the bridges is seen in the position-position-velocity diagram obtained using the SCOUSEPY algorithm. Star cluster formation with an over-density of Class I and Class II young stellar objects is also seen towards this cloud complex, likely triggered by the cloud collision event., Comment: 16 pages, 9 figures, 2 appendices, Accepted for publication in AJ

Humans use their knowledge of common house layouts obtained from previous experiences to predict nearby rooms while navigating in new environments. This greatly helps them navigate previously unseen environments and locate their target room. To provide layout prior knowledge to navigational agents based on common human living spaces, we propose WIN (\textit{W}hat \textit{I}s \textit{N}ear), a commonsense learning model for Vision Language Navigation (VLN) tasks. VLN requires an agent to traverse indoor environments based on descriptive navigational instructions. Unlike existing layout learning works, WIN predicts the local neighborhood map based on prior knowledge of living spaces and current observation, operating on an imagined global map of the entire environment. The model infers neighborhood regions based on visual cues of current observations, navigational history, and layout common sense. We show that local-global planning based on locality knowledge and predicting the indoor layout allows the agent to efficiently select the appropriate action. Specifically, we devised a cross-modal transformer that utilizes this locality prior for decision-making in addition to visual inputs and instructions. Experimental results show that locality learning using WIN provides better generalizability compared to classical VLN agents in unseen environments. Our model performs favorably on standard VLN metrics, with Success Rate 68\% and Success weighted by Path Length 63\% in unseen environments.

Modern micromanipulation techniques typically involve trapping using electromagnetic, acoustic or flow fields that produce stresses on the trapped particles thereby precluding stress-free manipulations. Here, we show that by employing polyhedral symmetries in a multichannel microfluidic design, we can separate the tasks of displacing and trapping a particle into two distinct sets of flow operations, each characterized and protected by their unique groups of symmetries. By combining only the displacing uniform flow modes to entrain and move targeted particles in arbitrary directions, we were able to realize symmetry-protected, stress-free micromanipulation in 3D. Furthermore, we engineered complex, microscale paths by programming and controlling the flow within each channel in real-time, resulting in multiple particles simultaneously following desired paths in the absence of any supervision or feedback. Our work therefore provides a general symmetry-group-based framework for understanding and engineering microfluidics and a novel platform for 3D stress-free manipulations.

Intracellular transport of cargoes in the cell is essential for the organization and functioning cells, especially those that are large and elongated. The cytoskeletal networks inside large cells can be highly complex, and this cytoskeletal organization can have impacts on the distance and trajectories of travel. Here, we experimentally created microtubule networks with varying mesh sizes and examined the ability of kinesin-driven quantum dot cargoes to traverse the network. Using the experimental data, we deduced parameters for cargo detachment at intersections and away from intersections, allowing us to create an analytical theory for the run length as a function of mesh size. We also used these parameters to perform simulations of cargoes along paths extracted from the experimental networks. We find excellent agreement between the trends in run length, displacement, and trajectory persistence length comparing the experimental and simulated trajectories., Comment: 23 Pages, 7 Figures, 4 Tables

Atmospheric escape is a fundamental process that affects the structure, composition, and evolution of many planets. The signatures of escape are detectable on close-in, gaseous exoplanets orbiting bright stars, owing to the high levels of extreme-ultraviolet irradiation from their parent stars. The Colorado Ultraviolet Transit Experiment (CUTE) is a CubeSat mission designed to take advantage of the near-ultraviolet stellar brightness distribution to conduct a survey of the extended atmospheres of nearby close-in planets. The CUTE payload is a magnifying NUV (2479~--~3306 Ang) spectrograph fed by a rectangular Cassegrain telescope (206mm x 84mm); the spectrogram is recorded on a back-illuminated, UV-enhanced CCD. The science payload is integrated into a 6U Blue Canyon Technology XB1 bus. CUTE was launched into a polar, low-Earth orbit on 27 September 2021 and has been conducting this transit spectroscopy survey following an on-orbit commissioning period. This paper presents the mission motivation, development path, and demonstrates the potential for small satellites to conduct this type of science by presenting initial on-orbit science observations. The primary science mission is being conducted in 2022~--~2023, with a publicly available data archive coming on line in 2023., Comment: 12 pages, 5 figures, AJ - accepted

A common type of cytoskeletal morphology involves multiple converging microbutubules with their minus ends collected and stabilized by a microtubule organizing center (MTOC) in the interior of the cell. This arrangement enables the ballistic transport of cargo bound to microtubules, both dynein mediated transport towards the MTOC and kinesin mediated transport away from it, interspersed with diffusion for unbound cargo-motor complexes. Spatial and temporal positioning of the MTOC allows for bidirectional transport towards and away from specific organelles and locations within the cell and also the sequestering and subsequent dispersal of dynein transported cargo. The general principles governing dynamics, efficiency and tunability of such transport in the MTOC vicinity is not fully understood. To address this, we develop a one-dimensional model that includes advective transport towards an attractor (such as the MTOC), and diffusive transport that allows particles to reach absorbing boundaries (such as cellular membranes). We calculated the mean first passage time (MFPT) for cargo to reach the boundaries as a measure of the effectiveness of sequestering (large MFPT) and diffusive dispersal (low MFPT). The MFPT experiences a dramatic growth in magnitude, transitioning from a low to high MFPT regime (dispersal to sequestering) over a window of cargo attachment/detachment rates that is close to in vivo values. We find that increasing either the attachment or detachment rate, while fixing the other, can result in optimal dispersal when the attractor is placed asymmetrically. Finally, we describe a rare event regime, where the escape location is exponentially sensitive to the attractor positioning. Our results suggest that structures such as the MTOC allow for the sensitive control of the spatial and temporal features of transport and corresponding function under physiological conditions.

Molecular clouds are sites of star formation. Magnetic fields are believed to play an important role in their dynamics and shaping morphology. We aim to study any possible correlation that might exist between the magnetic fields orientation inside the clouds and the magnetic fields at envelope scales and their connection with respect to the observed morphology of the selected clouds. We examine the magnetic field orientation towards the clouds L1512, L1523, L1333, L1521E, L1544, L1517, L1780, and L183 using optical and \textit{Planck} polarization observations. We also found the correlation between the ambient magnetic field and core orientations derived using \textit{Astrodendrogram} on the \textit{Herschel} 250 $\mu$m data. We find that the magnetic fields derived from optical and \textit{Planck} agree with each other. The derived magnetic fields are aligned along the observed emission of each cloud as seen in \textit{Herschel} 250 $\mu$m data. We also find that the relative orientation between the cores and the magnetic fields is random. This lack of correlation may arise due to the fact that the core orientation could also be influenced by the different magnetization within individual clouds at higher densities or the feedback effects which may vary from cloud to cloud. The estimated magnetic field strength and the mass-to-flux ratio suggest that all the clouds are in a magnetically critical state except L1333, L1521E, and L183 where the cloud envelope could be strongly supported by the magnetic field lines., Comment: 19 pages, 5 figures and 5 tables; Accepted for publication in MNRAS

This talk presents Gopcaml-mode, the first structural editing plugin for OCaml. We will give a tour of the main plugin features, discussing the plugin's internal design and its integration with existing OCaml and GNU Emacs toolchains., Comment: Presented at OCaml workshop at ICFP 2021