arXiv subjects

Stuart Anderson

Publications and source records attributed to Stuart Anderson.

14 recordsLinked to original sources

arXiv2024

Sapiens: Foundation for Human Vision Models

We present Sapiens, a family of models for four fundamental human-centric vision tasks -- 2D pose estimation, body-part segmentation, depth estimation, and surface normal prediction. Our models natively support 1K high-resolution inference and are extremely easy to adapt for individual tasks by simply fine-tuning models pretrained on over 300 million in-the-wild human images. We observe that, given the same computational budget, self-supervised pretraining on a curated dataset of human images significantly boosts the performance for a diverse set of human-centric tasks. The resulting models exhibit remarkable generalization to in-the-wild data, even when labeled data is scarce or entirely synthetic. Our simple model design also brings scalability -- model performance across tasks improves as we scale the number of parameters from 0.3 to 2 billion. Sapiens consistently surpasses existing baselines across various human-centric benchmarks. We achieve significant improvements over the prior state-of-the-art on Humans-5K (pose) by 7.6 mAP, Humans-2K (part-seg) by 17.1 mIoU, Hi4D (depth) by 22.4% relative RMSE, and THuman2 (normal) by 53.5% relative angular error. Project page: https://about.meta.com/realitylabs/codecavatars/sapiens.

cs.CV↗

arXiv2022

MidasTouch: Monte-Carlo inference over distributions across sliding touch

We present MidasTouch, a tactile perception system for online global localization of a vision-based touch sensor sliding on an object surface. This framework takes in posed tactile images over time, and outputs an evolving distribution of sensor pose on the object's surface, without the need for visual priors. Our key insight is to estimate local surface geometry with tactile sensing, learn a compact representation for it, and disambiguate these signals over a long time horizon. The backbone of MidasTouch is a Monte-Carlo particle filter, with a measurement model based on a tactile code network learned from tactile simulation. This network, inspired by LIDAR place recognition, compactly summarizes local surface geometries. These generated codes are efficiently compared against a precomputed tactile codebook per-object, to update the pose distribution. We further release the YCB-Slide dataset of real-world and simulated forceful sliding interactions between a vision-based tactile sensor and standard YCB objects. While single-touch localization can be inherently ambiguous, we can quickly localize our sensor by traversing salient surface geometries. Project page: https://suddhu.github.io/midastouch-tactile/

cs.RO↗

arXiv2022

Grasp Stability Prediction with Sim-to-Real Transfer from Tactile Sensing

Robot simulation has been an essential tool for data-driven manipulation tasks. However, most existing simulation frameworks lack either efficient and accurate models of physical interactions with tactile sensors or realistic tactile simulation. This makes the sim-to-real transfer for tactile-based manipulation tasks still challenging. In this work, we integrate simulation of robot dynamics and vision-based tactile sensors by modeling the physics of contact. This contact model uses simulated contact forces at the robot's end-effector to inform the generation of realistic tactile outputs. To eliminate the sim-to-real transfer gap, we calibrate our physics simulator of robot dynamics, contact model, and tactile optical simulator with real-world data, and then we demonstrate the effectiveness of our system on a zero-shot sim-to-real grasp stability prediction task where we achieve an average accuracy of 90.7% on various objects. Experiments reveal the potential of applying our simulation framework to more complicated manipulation tasks. We open-source our simulation framework at https://github.com/CMURoboTouch/Taxim/tree/taxim-robot.

cs.RO↗

arXiv2022

Theseus: A Library for Differentiable Nonlinear Optimization

We present Theseus, an efficient application-agnostic open source library for differentiable nonlinear least squares (DNLS) optimization built on PyTorch, providing a common framework for end-to-end structured learning in robotics and vision. Existing DNLS implementations are application specific and do not always incorporate many ingredients important for efficiency. Theseus is application-agnostic, as we illustrate with several example applications that are built using the same underlying differentiable components, such as second-order optimizers, standard costs functions, and Lie groups. For efficiency, Theseus incorporates support for sparse solvers, automatic vectorization, batching, GPU acceleration, and gradient computation with implicit differentiation and direct loss minimization. We do extensive performance evaluation in a set of applications, demonstrating significant efficiency gains and better scalability when these features are incorporated. Project page: https://sites.google.com/view/theseus-ai

cs.RO↗

arXiv2022

Vision Checklist: Towards Testable Error Analysis of Image Models to Help System Designers Interrogate Model Capabilities

Using large pre-trained models for image recognition tasks is becoming increasingly common owing to the well acknowledged success of recent models like vision transformers and other CNN-based models like VGG and Resnet. The high accuracy of these models on benchmark tasks has translated into their practical use across many domains including safety-critical applications like autonomous driving and medical diagnostics. Despite their widespread use, image models have been shown to be fragile to changes in the operating environment, bringing their robustness into question. There is an urgent need for methods that systematically characterise and quantify the capabilities of these models to help designers understand and provide guarantees about their safety and robustness. In this paper, we propose Vision Checklist, a framework aimed at interrogating the capabilities of a model in order to produce a report that can be used by a system designer for robustness evaluations. This framework proposes a set of perturbation operations that can be applied on the underlying data to generate test samples of different types. The perturbations reflect potential changes in operating environments, and interrogate various properties ranging from the strictly quantitative to more qualitative. Our framework is evaluated on multiple datasets like Tinyimagenet, CIFAR10, CIFAR100 and Camelyon17 and for models like ViT and Resnet. Our Vision Checklist proposes a specific set of evaluations that can be integrated into the previously proposed concept of a model card. Robustness evaluations like our checklist will be crucial in future safety evaluations of visual perception modules, and be useful for a wide range of stakeholders including designers, deployers, and regulators involved in the certification of these systems. Source code of Vision Checklist would be open for public use.

cs.CV↗

arXiv2021

PatchGraph: In-hand tactile tracking with learned surface normals

We address the problem of tracking 3D object poses from touch during in-hand manipulations. Specifically, we look at tracking small objects using vision-based tactile sensors that provide high-dimensional tactile image measurements at the point of contact. While prior work has relied on a-priori information about the object being localized, we remove this requirement. Our key insight is that an object is composed of several local surface patches, each informative enough to achieve reliable object tracking. Moreover, we can recover the geometry of this local patch online by extracting local surface normal information embedded in each tactile image. We propose a novel two-stage approach. First, we learn a mapping from tactile images to surface normals using an image translation network. Second, we use these surface normals within a factor graph to both reconstruct a local patch map and use it to infer 3D object poses. We demonstrate reliable object tracking for over $100$ contact sequences across unique shapes with four objects in simulation and two objects in the real-world. Supplementary video: https://youtu.be/FHks--haOGY

cs.RO↗

arXiv2021

LEO: Learning Energy-based Models in Factor Graph Optimization

We address the problem of learning observation models end-to-end for estimation. Robots operating in partially observable environments must infer latent states from multiple sensory inputs using observation models that capture the joint distribution between latent states and observations. This inference problem can be formulated as an objective over a graph that optimizes for the most likely sequence of states using all previous measurements. Prior work uses observation models that are either known a-priori or trained on surrogate losses independent of the graph optimizer. In this paper, we propose a method to directly optimize end-to-end tracking performance by learning observation models with the graph optimizer in the loop. This direct approach may appear, however, to require the inference algorithm to be fully differentiable, which many state-of-the-art graph optimizers are not. Our key insight is to instead formulate the problem as that of energy-based learning. We propose a novel approach, LEO, for learning observation models end-to-end with graph optimizers that may be non-differentiable. LEO alternates between sampling trajectories from the graph posterior and updating the model to match these samples to ground truth trajectories. We propose a way to generate such samples efficiently using incremental Gauss-Newton solvers. We compare LEO against baselines on datasets drawn from two distinct tasks: navigation and real-world planar pushing. We show that LEO is able to learn complex observation models with lower errors and fewer samples. Supplementary video: https://youtu.be/YqzlUPudfkA

cs.RO↗

arXiv2021

First demonstration of early warning gravitational wave alerts

Gravitational-wave observations became commonplace in Advanced LIGO-Virgo's recently concluded third observing run. 56 non-retracted candidates were identified and publicly announced in near real time. Gravitational waves from binary neutron star mergers, however, remain of special interest since they can be precursors to high-energy astrophysical phenomena like $\gamma$-ray bursts and kilonovae. While late-time electromagnetic emissions provide important information about the astrophysical processes within, the prompt emission along with gravitational waves uniquely reveals the extreme matter and gravity during - and in the seconds following - merger. Rapid communication of source location and properties from the gravitational-wave data is crucial to facilitate multi-messenger follow-up of such sources. This is especially enabled if the partner facilities are forewarned via an early-warning (pre-merger) alert. Here we describe the commissioning and performance of such a low-latency infrastructure within LIGO-Virgo. We present results from an end-to-end mock data challenge that detects binary neutron star mergers and alerts partner facilities before merger. We set expectations for these alerts in future observing runs.

astro-ph.HE↗

arXiv2020

Learning Tactile Models for Factor Graph-based Estimation

We're interested in the problem of estimating object states from touch during manipulation under occlusions. In this work, we address the problem of estimating object poses from touch during planar pushing. Vision-based tactile sensors provide rich, local image measurements at the point of contact. A single such measurement, however, contains limited information and multiple measurements are needed to infer latent object state. We solve this inference problem using a factor graph. In order to incorporate tactile measurements in the graph, we need local observation models that can map high-dimensional tactile images onto a low-dimensional state space. Prior work has used low-dimensional force measurements or engineered functions to interpret tactile measurements. These methods, however, can be brittle and difficult to scale across objects and sensors. Our key insight is to directly learn tactile observation models that predict the relative pose of the sensor given a pair of tactile images. These relative poses can then be incorporated as factors within a factor graph. We propose a two-stage approach: first we learn local tactile observation models supervised with ground truth data, and then integrate these models along with physics and geometric factors within a factor graph optimizer. We demonstrate reliable object tracking using only tactile feedback for 150 real-world planar pushing sequences with varying trajectories across three object shapes. Supplementary video: https://youtu.be/y1kBfSmi8w0

cs.RO↗

arXiv2019

The US Program in Ground-Based Gravitational Wave Science: Contribution from the LIGO Laboratory

Recent gravitational-wave observations from the LIGO and Virgo observatories have brought a sense of great excitement to scientists and citizens the world over. Since September 2015,10 binary black hole coalescences and one binary neutron star coalescence have been observed. They have provided remarkable, revolutionary insight into the "gravitational Universe" and have greatly extended the field of multi-messenger astronomy. At present, Advanced LIGO can see binary black hole coalescences out to redshift 0.6 and binary neutron star coalescences to redshift 0.05. This probes only a very small fraction of the volume of the observable Universe. However, current technologies can be extended to construct "$3^\mathrm{rd}$ Generation" (3G) gravitational-wave observatories that would extend our reach to the very edge of the observable Universe. The event rates over such a large volume would be in the hundreds of thousands per year (i.e.tens per hour). Such 3G detectors would have a 10-fold improvement in strain sensitivity over the current generation of instruments, yielding signal-to-noise ratios of 1000 for events like those already seen. Several concepts are being studied for which engineering studies and reliable cost estimates will be developed in the next 5 years.

astro-ph.IM↗

arXiv2017

BOSS-LDG: A Novel Computational Framework that Brings Together Blue Waters, Open Science Grid, Shifter and the LIGO Data Grid to Accelerate Gravitational Wave Discovery

We present a novel computational framework that connects Blue Waters, the NSF-supported, leadership-class supercomputer operated by NCSA, to the Laser Interferometer Gravitational-Wave Observatory (LIGO) Data Grid via Open Science Grid technology. To enable this computational infrastructure, we configured, for the first time, a LIGO Data Grid Tier-1 Center that can submit heterogeneous LIGO workflows using Open Science Grid facilities. In order to enable a seamless connection between the LIGO Data Grid and Blue Waters via Open Science Grid, we utilize Shifter to containerize LIGO's workflow software. This work represents the first time Open Science Grid, Shifter, and Blue Waters are unified to tackle a scientific problem and, in particular, it is the first time a framework of this nature is used in the context of large scale gravitational wave data analysis. This new framework has been used in the last several weeks of LIGO's second discovery campaign to run the most computationally demanding gravitational wave search workflows on Blue Waters, and accelerate discovery in the emergent field of gravitational wave astrophysics. We discuss the implications of this novel framework for a wider ecosystem of Higher Performance Computing users.

cs.DC↗

arXiv2017

Towards high-throughput 3D insect capture for species discovery and diagnostics

Digitisation of natural history collections not only preserves precious information about biological diversity, it also enables us to share, analyse, annotate and compare specimens to gain new insights. High-resolution, full-colour 3D capture of biological specimens yields color and geometry information complementary to other techniques (e.g., 2D capture, electron scanning and micro computed tomography). However 3D colour capture of small specimens is slow for reasons including specimen handling, the narrow depth of field of high magnification optics, and the large number of images required to resolve complex shapes of specimens. In this paper, we outline techniques to accelerate 3D image capture, including using a desktop robotic arm to automate the insect handling process; using a calibrated pan-tilt rig to avoid attaching calibration targets to specimens; using light field cameras to capture images at an extended depth of field in one shot; and using 3D Web and mixed reality tools to facilitate the annotation, distribution and visualisation of 3D digital models.

cs.CV↗

arXiv2012

Theoretical And Technological Building Blocks For An Innovation Accelerator

The scientific system that we use today was devised centuries ago and is inadequate for our current ICT-based society: the peer review system encourages conservatism, journal publications are monolithic and slow, data is often not available to other scientists, and the independent validation of results is limited. Building on the Innovation Accelerator paper by Helbing and Balietti (2011) this paper takes the initial global vision and reviews the theoretical and technological building blocks that can be used for implementing an innovation (in first place: science) accelerator platform driven by re-imagining the science system. The envisioned platform would rest on four pillars: (i) Redesign the incentive scheme to reduce behavior such as conservatism, herding and hyping; (ii) Advance scientific publications by breaking up the monolithic paper unit and introducing other building blocks such as data, tools, experiment workflows, resources; (iii) Use machine readable semantics for publications, debate structures, provenance etc. in order to include the computer as a partner in the scientific process, and (iv) Build an online platform for collaboration, including a network of trust and reputation among the different types of stakeholders in the scientific system: scientists, educators, funding agencies, policy makers, students and industrial innovators among others. Any such improvements to the scientific system must support the entire scientific process (unlike current tools that chop up the scientific process into disconnected pieces), must facilitate and encourage collaboration and interdisciplinarity (again unlike current tools), must facilitate the inclusion of intelligent computing in the scientific process, must facilitate not only the core scientific process, but also accommodate other stakeholders such science policy makers, industrial innovators, and the general public.

cs.DL↗

arXiv2003

The Arecibo 430-MHz Intermediate Galactic Latitude Survey: Discovery of Nine Radio Pulsars

We have used the Arecibo Radio Telescope to search for millisecond pulsars in two intermediate Galactic latitude regions (7 deg < | b | < 20 deg) accessible to this telescope. For these latitudes the useful millisecond pulsar search volume achieved by Arecibo's 430-MHz beam is predicted to be maximal. Searching a total of 130 square degrees, we have discovered nine new pulsars and detected four previously known objects. We compare the results of this survey with those of other 430-MHz surveys carried out at Arecibo and of an intermediate latitude survey made at Parkes that included part of our search area; the latter independently found two of the nine pulsars we have discovered. At least six of our discoveries are isolated pulsars with ages between 5 and 300 Myr; one of these, PSR J1819+1305, exhibits very marked and periodic nulling. We have also found a recycled pulsar, PSR J2016+1948. With a rotational period of 65 ms, this is a member of a binary system with a 635-day orbital period. We discuss some of the the properties of this system in detail, and indicate its potential to provide a test of the Strong Equivalence Principle. This pulsar and PSR J0407+16, a similar system now being timed at Arecibo, are by far the best systems known for such a test.

astro-ph↗