arXiv subjects

David Matthews

Publications and source records attributed to David Matthews.

14 recordsLinked to original sources

arXiv2026

Computational Design of a Low-Visibility UAV Using a Human-Aligned Perceptual Metric

We introduce Phantom Twist, a type of single-propeller UAV designed to achieve low visibility through high-speed spinning and the exploitation of motion blur. We develop a two-stage automated design pipeline that optimizes the placement of functional components including batteries, control PCB, motor-propeller assembly, and counterweights. The pipeline minimizes visibility as measured by a human-aligned perceptual metric (LPIPS) while strictly satisfying inertial and aerodynamic constraints required for stable flight. We validate this approach through fabrication and flight testing of multiple prototypes. These tests confirm that our pipeline produces stable, controllable designs and that the optimized UAV exhibits significantly reduced visual perceptibility compared to conventional quadcopters.

cs.RO↗

arXiv2025

Agile legged locomotion in reconfigurable modular robots

Legged machines are becoming increasingly agile and adaptive but they have so far lacked the morphological diversity of legged animals, which have been rearranged and reshaped to fill millions of niches. Unlike their biological counterparts, legged machines have largely converged over the past decade to canonical quadrupedal and bipedal architectures that cannot be easily reconfigured to meet new tasks or recover from injury. Here we introduce autonomous modular legs: agile yet minimal, single-degree-of-freedom jointed links that can learn complex dynamic behaviors and may be freely attached to form multilegged machines at the meter scale. This enables rapid repair, redesign, and recombination of highly-dynamic modular agents that move quickly and acrobatically (non-quasistatically) through unstructured environments. Because each module is itself a complete agent, the bodies that contain them can sustain deep structural damage that would completely disable other legged robots. We also show how to encode the vast space of possible body configurations into a compact latent design space that can be efficiently explored, revealing a wide diversity of novel legged forms.

cs.RO↗

arXiv2025

An Updated Detection Pipeline for Precursor Emission in Type II Supernova 2020tlf

We present a new photometric pipeline for the detection of pre-supernova (pre-SN) emission in the Young Supernova Experiment (YSE) sky survey. The method described is applied to SN 2020tlf, a type II SN (SN II) with precursor emission in the last ~100 days before first light. We re-analyze the YSE griz-band light curves of SN 2020tlf and provide revised pre-explosion photometry that includes a robust list of confident detection and limiting magnitudes. Compared to the results of Jacobson-Galan et al. 2022a, this new analysis yields fewer total r/i/z-band pre-SN detections at phases > -100 days. Furthermore, we discourage the use of the blackbody modeling of the pre-explosion spectral energy distribution, the pre-SN bolometric light curve and the blackbody model parameters presented in Jacobson-Galan et al. 2022a. Nevertheless, binned photometry of SN 2020tlf confirms a consistent progenitor luminosity of ~10$^{40}$ erg s$^{-1}$ before explosion.

astro-ph.HE↗

arXiv2024

Dinosaur in a Haystack : X-ray View of the Entrails of SN 2023ixf and the Radio Afterglow of Its Interaction with the Medium Spawned by the Progenitor Star (Paper 1)

We present the results from our extensive hard-to-soft X-ray (NuSTAR, Swift-XRT, XMM-Newton, Chandra) and meter-to-mm wave radio (GMRT, VLA, NOEMA) monitoring campaign of the very nearby (d $=6.9$ Mpc) Type II SN2023ixf spanning $\approx$ 4--165 d post-explosion. This unprecedented dataset enables inferences on the explosion's circumstellar medium (CSM) density and geometry. Specifically, we find that the luminous X-ray emission is well modeled by thermal free-free radiation from the forward shock with rapidly decreasing photo-electric absorption with time. The radio spectrum is dominated by synchrotron radiation from the same shock, and the NOEMA detection of high-frequency radio emission may indicate a new component consistent with the secondary origin. Similar to the X-rays, the level of free-free absorption affecting the radio spectrum rapidly decreases with time as a consequence of the shock propagation into the dense CSM. While the X-ray and the radio modeling independently support the presence of a dense medium corresponding to an \emph{effective} mass-loss rate $\dot{M} \approx 10^{-4}\, \rm M_{\odot}\,yr^{-1}$ at $R = (0.4-14) \times 10^{15}$ (for $v_{\rm w}=\rm 25 \,km\,s^{-1}$), our study points at a complex CSM density structure with asymmetries and clumps. The inferred densities are $\approx$10--100 times those of typical red supergiants, indicating an extreme mass-loss phase of the progenitor in the $\approx$200 years preceding core collapse, which leads to the most X-ray luminous Type II SN and the one with the most delayed emergence of radio emission. These results add to the picture of the complex mass-loss history of massive stars on the verge of collapse and demonstrate the need for panchromatic campaigns to fully map their intricate environments.

astro-ph.HE↗

arXiv2024

PS1-11aop: Probing the Mass Loss History of a Luminous Interacting Supernova Prior to its Final Eruption with Multi-wavelength Observations

Luminous interacting supernovae are a class of stellar explosions whose progenitors underwent vigorous mass loss in the years prior to core-collapse. While the mechanism by which this material is ejected is still debated, obtaining the full density profile of the circumstellar medium (CSM) could reveal more about this process. Here, we present an extensive multi-wavelength study of PS1-11aop, a luminous and slowly declining Type IIn SN discovered by the PanSTARRS Medium Deep Survey. PS1-11aop had a peak r-band magnitude of $-$20.5\,mag, a total radiated energy $>$ 8$\times$10$^{50}$\,erg, and it exploded near the center of a star-forming galaxy with super-solar metallicity. We obtained multiple detections at the location of PS1-11aop in the radio and X-ray bands between 4 and 10\,years post-explosion, and if due to the SN, it is one of the most luminous radio supernovae identified to date. Taken together, the multiwavelength properties of PS1-11aop are consistent with a CSM density profile with multiple zones. The early optical emission is consistent with the supernova blastwave interacting with a dense and confined CSM shell which contains multiple solar masses of material that was likely ejected in the final $<$10-100 years prior to the explosion,($\sim$0.05$-$1.0 M$_{\odot}$yr$^{-1}$ at radii of $\lesssim$10$^{16}$\,cm). The radio observations, on the other hand, are consistent with a sparser environment ($\lesssim$2$\times 10^{-3}$ M$_{\odot}$yr$^{-1}$ at radii of $\sim$0.5-1$\times$10$^{17}$\,cm) -- thus probing the history of the progenitor star prior to its final mass loss episode.

astro-ph.HE↗

arXiv2024

Evolution and learning in differentiable robots

The automatic design of robots has existed for 30 years but has been constricted by serial non-differentiable design evaluations, premature convergence to simple bodies or clumsy behaviors, and a lack of sim2real transfer to physical machines. Thus, here we employ massively-parallel differentiable simulations to rapidly and simultaneously optimize individual neural control of behavior across a large population of candidate body plans and return a fitness score for each design based on the performance of its fully optimized behavior. Non-differentiable changes to the mechanical structure of each robot in the population -- mutations that rearrange, combine, add, or remove body parts -- were applied by a genetic algorithm in an outer loop of search, generating a continuous flow of novel morphologies with highly-coordinated and graceful behaviors honed by gradient descent. This enabled the exploration of several orders-of-magnitude more designs than all previous methods, despite the fact that robots here have the potential to be much more complex, in terms of number of independent motors, than those in prior studies. We found that evolution reliably produces ``increasingly differentiable'' robots: body plans that smooth the loss landscape in which learning operates and thereby provide better training paths toward performant behaviors. Finally, one of the highly differentiable morphologies discovered in simulation was realized as a physical robot and shown to retain its optimized behavior. This provides a cyberphysical platform to investigate the relationship between evolution and learning in biological systems and broadens our understanding of how a robot's physical structure can influence the ability to train policies for it. Videos and code at https://sites.google.com/view/eldir.

cs.RO↗

arXiv2023

Reinforcement learning for freeform robot design

Inspired by the necessity of morphological adaptation in animals, a growing body of work has attempted to expand robot training to encompass physical aspects of a robot's design. However, reinforcement learning methods capable of optimizing the 3D morphology of a robot have been restricted to reorienting or resizing the limbs of a predetermined and static topological genus. Here we show policy gradients for designing freeform robots with arbitrary external and internal structure. This is achieved through actions that deposit or remove bundles of atomic building blocks to form higher-level nonparametric macrostructures such as appendages, organs and cavities. Although results are provided for open loop control only, we discuss how this method could be adapted for closed loop control and sim2real transfer to physical machines in future.

cs.RO↗

arXiv2023

Efficient automatic design of robots

Robots are notoriously difficult to design because of complex interdependencies between their physical structure, sensory and motor layouts, and behavior. Despite this, almost every detail of every robot built to date has been manually determined by a human designer after several months or years of iterative ideation, prototyping, and testing. Inspired by evolutionary design in nature, the automated design of robots using evolutionary algorithms has been attempted for two decades, but it too remains inefficient: days of supercomputing are required to design robots in simulation that, when manufactured, exhibit desired behavior. Here we show for the first time de-novo optimization of a robot's structure to exhibit a desired behavior, within seconds on a single consumer-grade computer, and the manufactured robot's retention of that behavior. Unlike other gradient-based robot design methods, this algorithm does not presuppose any particular anatomical form; starting instead from a randomly-generated apodous body plan, it consistently discovers legged locomotion, the most efficient known form of terrestrial movement. If combined with automated fabrication and scaled up to more challenging tasks, this advance promises near instantaneous design, manufacture, and deployment of unique and useful machines for medical, environmental, vehicular, and space-based tasks.

cs.RO↗

arXiv2023

Scary Barbie: An Extremely Energetic, Long-Duration Tidal Disruption Event Candidate Without a Detected Host Galaxy at z = 0.995

We report multi-wavelength observations and characterization of the ultraluminous transient AT 2021lwx (ZTF20abrbeie; aka ``Barbie'') identified in the alert stream of the Zwicky Transient Facility (ZTF) using a Recommender Engine For Intelligent Transient Tracking (REFITT) filter on the ANTARES alert broker. From a spectroscopically measured redshift of 0.995, we estimate a peak observed pseudo-bolometric luminosity of log (L$_{\text{max}} / [\text{erg}/\text{s}]$) = 45.7 from slowly fading ztf-$\it{g}$ and ztf-$r$ light curves spanning over 1000 observer-frame days. The host galaxy is not detected in archival Pan-STARRS observations ($g > 23.3$ mag), implying a lower limit to the outburst amplitude of more than 5 mag relative to the quiescent host galaxy. Optical spectra from Lick and Keck Observatories exhibit strong emission lines with narrow cores from the H Balmer series and ultraviolet semi-forbidden lines of Si III] $\lambda$1892, C III] $\lambda$1909, and C II] $\lambda$2325. Typical nebular lines in AGN spectra from ions such as [O II] and [O III] are not detected. These spectral features, along with the smooth light curve that is unlike most AGN flaring activity, and the luminosity that exceeds any observed or theorized supernova, lead us to conclude that AT 2021lwx is most likely an extreme tidal disruption event (TDE). Modeling of ZTF photometry with MOSFiT suggests that the TDE was between a $\approx 14 M_{\odot}$ star and a supermassive black hole of mass $M_{\text{BH}} \sim$ $10^{8} M_{\odot}$. Continued monitoring of the still-evolving light curve along with deep imaging of the field once AT 2021lwx has faded can test this hypothesis and potentially detect the host galaxy.

astro-ph.HE↗

arXiv2022

Evidence for Extended Hydrogen-Poor CSM in the Three-Peaked Light Curve of Stripped Envelope Ib Supernova

We present multi-band ATLAS photometry for SN 2019tsf, a stripped-envelope Type Ib supernova (SESN). The SN shows a triple-peaked light curve and a late (re-)brightening, making it unique among stripped-envelope systems. The re-brightening observations represent the latest photometric measurements of a multi-peaked Type Ib SN to date. As late-time photometry and spectroscopy suggest no hydrogen, the potential circumstellar material (CSM) must be H-poor. Moreover, late (>150 days) spectra show no signs of narrow emission lines, further disfavouring CSM interaction. On the contrary, an extended CSM structure is seen through a follow-up radio campaign with Karl G. Jansky Very Large Array (VLA), indicating a source of bright optically thick radio emission at late times, which is highly unusual among H-poor SESNe. We attribute this phenomenology to an interaction of the supernova ejecta with spherically-asymmetric CSM, potentially disk-like, and we present several models that can potentially explain the origin of this rare Type Ib supernova. The warped disc model paints a novel picture, where the tertiary companion perturbs the progenitors CSM, that can explain the multi-peaked light curves of SNe, and here we apply it to SN 2019tsf. This SN 2019tsf is likely a member of a new sub-class of Type Ib SNe and among the recently discovered class of SNe that undergo mass transfer at the moment of explosion

astro-ph.HE↗

arXiv2021

Radio and X-ray observations of the luminous Fast Blue Optical Transient AT2020xnd

We present deep X-ray and radio observations of the Fast Blue Optical Transient (FBOT) AT2020xnd/ZTF20acigmel at $z=0.2433$ from $13$d to $269$d after explosion. AT2020xnd belongs to the category of optically luminous FBOTs with similarities to the archetypal event AT2018cow. AT2020xnd shows luminous radio emission reaching $L_{\nu}\approx8\times10^{29}$ergs$^{-1}$Hz$^{-1}$ at 20GHz and $75$d post explosion, accompanied by luminous and rapidly fading soft X-ray emission peaking at $L_{X}\approx6\times10^{42}$ergs$^{-1}$. Interpreting the radio emission in the context of synchrotron radiation from the explosion's shock interaction with the environment we find that AT2020xnd launched a high-velocity outflow ($v\sim$0.1-0.2$c$) propagating into a dense circumstellar medium (effective $\dot M\approx10^{-3}M_{\rm{sol}}$yr$^{-1}$ for an assumed wind velocity of $v_w=1000$kms$^{-1}$). Similar to AT2018cow, the detected X-ray emission is in excess compared to the extrapolated synchrotron spectrum and constitutes a different emission component, possibly powered by accretion onto a newly formed black hole or neutron star. These properties make AT2020xnd a high-redshift analog to AT2018cow, and establish AT2020xnd as the fourth member of the class of optically-luminous FBOTs with luminous multi-wavelength counterparts.

astro-ph.HE↗

arXiv2021

The Role of Cross-Silo Federated Learning in Facilitating Data Sharing in the Agri-Food Sector

Data sharing remains a major hindering factor when it comes to adopting emerging AI technologies in general, but particularly in the agri-food sector. Protectiveness of data is natural in this setting; data is a precious commodity for data owners, which if used properly can provide them with useful insights on operations and processes leading to a competitive advantage. Unfortunately, novel AI technologies often require large amounts of training data in order to perform well, something that in many scenarios is unrealistic. However, recent machine learning advances, e.g. federated learning and privacy-preserving technologies, can offer a solution to this issue via providing the infrastructure and underpinning technologies needed to use data from various sources to train models without ever sharing the raw data themselves. In this paper, we propose a technical solution based on federated learning that uses decentralized data, (i.e. data that are not exchanged or shared but remain with the owners) to develop a cross-silo machine learning model that facilitates data sharing across supply chains. We focus our data sharing proposition on improving production optimization through soybean yield prediction, and provide potential use-cases that such methods can assist in other problem settings. Our results demonstrate that our approach not only performs better than each of the models trained on an individual data source, but also that data sharing in the agri-food sector can be enabled via alternatives to data exchange, whilst also helping to adopt emerging machine learning technologies to boost productivity.

cs.LG↗

arXiv2019

Word2vec to behavior: morphology facilitates the grounding of language in machines

Enabling machines to respond appropriately to natural language commands could greatly expand the number of people to whom they could be of service. Recently, advances in neural network-trained word embeddings have empowered non-embodied text-processing algorithms, and suggest they could be of similar utility for embodied machines. Here we introduce a method that does so by training robots to act similarly to semantically-similar word2vec encoded commands. We show that this enables them to act appropriately, after training, to previously-unheard commands. Finally, we show that inducing such an alignment between motoric and linguistic similarities can be facilitated or hindered by the mechanical structure of the robot. This points to future, large scale methods that find and exploit relationships between action, language, and robot structure.

cs.CL↗

arXiv2018

Learning Cheap and Novel Flight Itineraries

We consider the problem of efficiently constructing cheap and novel round trip flight itineraries by combining legs from different airlines. We analyse the factors that contribute towards the price of such itineraries and find that many result from the combination of just 30% of airlines and that the closer the departure of such itineraries is to the user's search date the more likely they are to be cheaper than the tickets from one airline. We use these insights to formulate the problem as a trade-off between the recall of cheap itinerary constructions and the costs associated with building them. We propose a supervised learning solution with location embeddings which achieves an AUC=80.48, a substantial improvement over simpler baselines. We discuss various practical considerations for dealing with the staleness and the stability of the model and present the design of the machine learning pipeline. Finally, we present an analysis of the model's performance in production and its impact on Skyscanner's users.

cs.LG↗