-
Winning Through Simplicity: Autonomous Car Design for Formula Student
Authors:
Tobias Friedrich,
Marco Müller,
Adrian Bauske,
Simon Härtl,
Johannes Herrmann,
David Förster,
Tobias Tietze,
Sebastian Sartor
Abstract:
This paper presents the design of an autonomous race car that is self-designed, self-developed, and self-built by the Elefant Racing team at the University of Bayreuth. The system is created to compete in the Formula Student Driverless competition. Its primary focus is on the Acceleration track, a straight 75-meter-long course, and the Skidpad track, which comprises two circles forming an eight. A…
▽ More
This paper presents the design of an autonomous race car that is self-designed, self-developed, and self-built by the Elefant Racing team at the University of Bayreuth. The system is created to compete in the Formula Student Driverless competition. Its primary focus is on the Acceleration track, a straight 75-meter-long course, and the Skidpad track, which comprises two circles forming an eight. Additionally, it is experimentally capable of competing in the Autocross and Trackdrive events, which feature tracks with previously unknown straights and curves. The paper details the hardware, software and sensor setup employed during the 2020/2021 season. Despite being developed by a small team with limited computer science expertise, the design won the Formula Student East Engineering Design award. Emphasizing simplicity and efficiency, the team employed streamlined techniques to achieve their success.
△ Less
Submitted 19 June, 2024;
originally announced June 2024.
-
MatFusion: A Generative Diffusion Model for SVBRDF Capture
Authors:
Sam Sartor,
Pieter Peers
Abstract:
We formulate SVBRDF estimation from photographs as a diffusion task. To model the distribution of spatially varying materials, we first train a novel unconditional SVBRDF diffusion backbone model on a large set of 312,165 synthetic spatially varying material exemplars. This SVBRDF diffusion backbone model, named MatFusion, can then serve as a basis for refining a conditional diffusion model to est…
▽ More
We formulate SVBRDF estimation from photographs as a diffusion task. To model the distribution of spatially varying materials, we first train a novel unconditional SVBRDF diffusion backbone model on a large set of 312,165 synthetic spatially varying material exemplars. This SVBRDF diffusion backbone model, named MatFusion, can then serve as a basis for refining a conditional diffusion model to estimate the material properties from a photograph under controlled or uncontrolled lighting. Our backbone MatFusion model is trained using only a loss on the reflectance properties, and therefore refinement can be paired with more expensive rendering methods without the need for backpropagation during training. Because the conditional SVBRDF diffusion models are generative, we can synthesize multiple SVBRDF estimates from the same input photograph from which the user can select the one that best matches the users' expectation. We demonstrate the flexibility of our method by refining different SVBRDF diffusion models conditioned on different types of incident lighting, and show that for a single photograph under colocated flash lighting our method achieves equal or better accuracy than existing SVBRDF estimation methods.
△ Less
Submitted 23 April, 2024;
originally announced June 2024.
-
Neural Scaling Laws for Embodied AI
Authors:
Sebastian Sartor,
Neil Thompson
Abstract:
Scaling laws have driven remarkable progress across machine learning domains like language modeling and computer vision. However, the exploration of scaling laws in embodied AI and robotics has been limited, despite the rapidly increasing usage of machine learning in this field. This paper presents the first study to quantify scaling laws for Robot Foundation Models (RFMs) and the use of LLMs in r…
▽ More
Scaling laws have driven remarkable progress across machine learning domains like language modeling and computer vision. However, the exploration of scaling laws in embodied AI and robotics has been limited, despite the rapidly increasing usage of machine learning in this field. This paper presents the first study to quantify scaling laws for Robot Foundation Models (RFMs) and the use of LLMs in robotics tasks. Through a meta-analysis spanning 198 research papers, we analyze how key factors like compute, model size, and training data quantity impact model performance across various robotic tasks. Our findings confirm that scaling laws apply to both RFMs and LLMs in robotics, with performance consistently improving as resources increase. The power law coefficients for RFMs closely match those of LLMs in robotics, resembling those found in computer vision and outperforming those for LLMs in the language domain. We also note that these coefficients vary with task complexity, with familiar tasks scaling more efficiently than unfamiliar ones, emphasizing the need for large and diverse datasets. Furthermore, we highlight the absence of standardized benchmarks in embodied AI. Most studies indicate diminishing returns, suggesting that significant resources are necessary to achieve high performance, posing challenges due to data and computational limitations. Finally, as models scale, we observe the emergence of new capabilities, particularly related to data and model size.
△ Less
Submitted 22 May, 2024;
originally announced May 2024.
-
Intrinsic Mesh Simplification
Authors:
Randy Shoemaker,
Sam Sartor,
Pieter Peers
Abstract:
This paper presents a novel simplification method for removing vertices from an intrinsic triangulation corresponding to extrinsic vertices lying on near-developable (i.e., with limited Gaussian curvature) and general surfaces. We greedily process all intrinsic vertices with an absolute Gaussian curvature below a user selected threshold. For each vertex, we repeatedly perform local intrinsic edge…
▽ More
This paper presents a novel simplification method for removing vertices from an intrinsic triangulation corresponding to extrinsic vertices lying on near-developable (i.e., with limited Gaussian curvature) and general surfaces. We greedily process all intrinsic vertices with an absolute Gaussian curvature below a user selected threshold. For each vertex, we repeatedly perform local intrinsic edge flips until the vertex reaches the desired valence (three for internal vertices or two for boundary vertices) such that removal of the vertex and incident edges can be locally performed in the intrinsic triangulation. Each removed vertex's intrinsic location is tracked via (intrinsic) barycentric coordinates that are updated to reflect changes in the intrinsic triangulation. We demonstrate the robustness and effectiveness of our method on the Thingi10k dataset and analyze the effect of the curvature threshold on the solutions of PDEs.
△ Less
Submitted 13 July, 2023;
originally announced July 2023.
-
The large scale polarization explorer (LSPE) for CMB measurements: performance forecast
Authors:
The LSPE collaboration,
G. Addamo,
P. A. R. Ade,
C. Baccigalupi,
A. M. Baldini,
P. M. Battaglia,
E. S. Battistelli,
A. Baù,
P. de Bernardis,
M. Bersanelli,
M. Biasotti,
A. Boscaleri,
B. Caccianiga,
S. Caprioli,
F. Cavaliere,
F. Cei,
K. A. Cleary,
F. Columbro,
G. Coppi,
A. Coppolecchia,
F. Cuttaia,
G. D'Alessandro,
G. De Gasperis,
M. De Petris,
V. Fafone
, et al. (80 additional authors not shown)
Abstract:
[Abridged] The measurement of the polarization of the Cosmic Microwave Background radiation is one of the current frontiers in cosmology. In particular, the detection of the primordial B-modes, could reveal the presence of gravitational waves in the early Universe. The detection of such component is at the moment the most promising technique to probe the inflationary theory describing the very ear…
▽ More
[Abridged] The measurement of the polarization of the Cosmic Microwave Background radiation is one of the current frontiers in cosmology. In particular, the detection of the primordial B-modes, could reveal the presence of gravitational waves in the early Universe. The detection of such component is at the moment the most promising technique to probe the inflationary theory describing the very early evolution of the Universe. We present the updated performance forecast of the Large Scale Polarization Explorer (LSPE), a program dedicated to the measurement of the CMB polarization. LSPE is composed of two instruments: Strip, a radiometer-based telescope on the ground in Tenerife, and SWIPE (Short-Wavelength Instrument for the Polarization Explorer) a bolometer-based instrument designed to fly on a winter arctic stratospheric long-duration balloon. The program is among the few dedicated to observation of the Northern Hemisphere, while most of the international effort is focused into ground-based observation in the Southern Hemisphere. Measurements are currently scheduled in Winter 2021/22 for SWIPE, with a flight duration up to 15 days, and in Summer 2021 with two years observations for Strip. We describe the main features of the two instruments, identifying the most critical aspects of the design, in terms of impact into performance forecast. We estimate the expected sensitivity of each instrument and propagate their combined observing power to the sensitivity to cosmological parameters, including the effect of scanning strategy, component separation, residual foregrounds and partial sky coverage. We also set requirements on the control of the most critical systematic effects and describe techniques to mitigate their impact. LSPE can reach a sensitivity in tensor-to-scalar ratio of $σ_r<0.01$, and improve constrains on other cosmological parameters.
△ Less
Submitted 9 August, 2021; v1 submitted 25 August, 2020;
originally announced August 2020.
-
Progress report on the Large Scale Polarization Explorer
Authors:
L. Lamagna,
G. Addamo,
P. A. R. Ade,
C. Baccigalupi,
A. M. Baldini,
P. M. Battaglia,
E. Battistelli,
A. Baù,
M. Bersanelli,
M. Biasotti,
C. Boragno,
A. Boscaleri,
B. Caccianiga,
S. Caprioli,
F. Cavaliere,
F. Cei,
K. A. Cleary,
F. Columbro,
G. Coppi,
A. Coppolecchia,
D. Corsini,
F. Cuttaia,
G. D'Alessandro,
P. de Bernardis,
G. De Gasperis
, et al. (74 additional authors not shown)
Abstract:
The Large Scale Polarization Explorer (LSPE) is a cosmology program for the measurement of large scale curl-like features (B-modes) in the polarization of the Cosmic Microwave Background. Its goal is to constrain the background of inflationary gravity waves traveling through the universe at the time of matter-radiation decoupling. The two instruments of LSPE are meant to synergically operate by co…
▽ More
The Large Scale Polarization Explorer (LSPE) is a cosmology program for the measurement of large scale curl-like features (B-modes) in the polarization of the Cosmic Microwave Background. Its goal is to constrain the background of inflationary gravity waves traveling through the universe at the time of matter-radiation decoupling. The two instruments of LSPE are meant to synergically operate by covering a large portion of the northern microwave sky. LSPE/STRIP is a coherent array of receivers planned to be operated from the Teide Observatory in Tenerife, for the control and characterization of the low-frequency polarized signals of galactic origin; LSPE/SWIPE is a balloon-borne bolometric polarimeter based on 330 large throughput multi-moded detectors, designed to measure the CMB polarization at 150 GHz and to monitor the polarized emission by galactic dust above 200 GHz. The combined performance and the expected level of systematics mitigation will allow LSPE to constrain primordial B-modes down to a tensor/scalar ratio of $10^{-2}$. We here report the status of the STRIP pre-commissioning phase and the progress in the characterization of the key subsystems of the SWIPE payload (namely the cryogenic polarization modulation unit and the multi-moded TES pixels) prior to receiver integration.
△ Less
Submitted 5 May, 2020; v1 submitted 3 May, 2020;
originally announced May 2020.
-
The STRIP instrument of the Large Scale Polarization Explorer: microwave eyes to map the Galactic polarized foregrounds
Authors:
C. Franceschet,
S. Realini,
A. Mennella,
G. Addamo,
A. Baù,
P. M. Battaglia,
M. Bersanelli,
B. Caccianiga,
S. Caprioli,
F. Cavaliere,
K. A. Cleary,
F. Cuttaia,
F. Del Torto,
V. Fafone,
Z. Farooqui,
R. T. Génova Santos,
T. C. Gaier,
M. Gervasi,
T. Ghigna,
F. Incardona,
S. Iovenitti,
M. Jones,
P. Kangaslahti,
R. Mainini,
D. Maino
, et al. (26 additional authors not shown)
Abstract:
In this paper we discuss the latest developments of the STRIP instrument of the "Large Scale Polarization Explorer" (LSPE) experiment. LSPE is a novel project that combines ground-based (STRIP) and balloon-borne (SWIPE) polarization measurements of the microwave sky on large angular scales to attempt a detection of the "B-modes" of the Cosmic Microwave Background polarization. STRIP will observe a…
▽ More
In this paper we discuss the latest developments of the STRIP instrument of the "Large Scale Polarization Explorer" (LSPE) experiment. LSPE is a novel project that combines ground-based (STRIP) and balloon-borne (SWIPE) polarization measurements of the microwave sky on large angular scales to attempt a detection of the "B-modes" of the Cosmic Microwave Background polarization. STRIP will observe approximately 25% of the Northern sky from the "Observatorio del Teide" in Tenerife, using an array of forty-nine coherent polarimeters at 43 GHz, coupled to a 1.5 m fully rotating crossed-Dragone telescope. A second frequency channel with six-elements at 95 GHz will be exploited as an atmospheric monitor. At present, most of the hardware of the STRIP instrument has been developed and tested at sub-system level. System-level characterization, starting in July 2018, will lead STRIP to be shipped and installed at the observation site within the end of the year. The on-site verification and calibration of the whole instrument will prepare STRIP for a 2-years campaign for the observation of the CMB polarization.
△ Less
Submitted 10 December, 2018;
originally announced December 2018.
-
DAS: a data management system for instrument tests and operations
Authors:
Marco Frailis,
Stefano Sartor,
Andrea Zacchei,
Marcello Lodi,
Roberto Cirami,
Fabio Pasian,
Massimo Trifoglio,
Andrea Bulgarelli,
Fulvio Gianotti,
Enrico Franceschi,
Luciano Nicastro,
Vito Conforti,
Andrea Zoli,
Ricky Smart,
Roberto Morbidelli,
Mauro Dadina
Abstract:
The Data Access System (DAS) is a metadata and data management software system, providing a reusable solution for the storage of data acquired both from telescopes and auxiliary data sources during the instrument development phases and operations. It is part of the Customizable Instrument WorkStation system (CIWS-FW), a framework for the storage, processing and quick-look at the data acquired from…
▽ More
The Data Access System (DAS) is a metadata and data management software system, providing a reusable solution for the storage of data acquired both from telescopes and auxiliary data sources during the instrument development phases and operations. It is part of the Customizable Instrument WorkStation system (CIWS-FW), a framework for the storage, processing and quick-look at the data acquired from scientific instruments. The DAS provides a data access layer mainly targeted to software applications: quick-look displays, pre-processing pipelines and scientific workflows. It is logically organized in three main components: an intuitive and compact Data Definition Language (DAS DDL) in XML format, aimed for user-defined data types; an Application Programming Interface (DAS API), automatically adding classes and methods supporting the DDL data types, and providing an object-oriented query language; a data management component, which maps the metadata of the DDL data types in a relational Data Base Management System (DBMS), and stores the data in a shared (network) file system. With the DAS DDL, developers define the data model for a particular project, specifying for each data type the metadata attributes, the data format and layout (if applicable), and named references to related or aggregated data types. Together with the DDL user-defined data types, the DAS API acts as the only interface to store, query and retrieve the metadata and data in the DAS system, providing both an abstract interface and a data model specific one in C, C++ and Python. The map** of metadata in the back-end database is automatic and supports several relational DBMSs, including MySQL, Oracle and PostgreSQL.
△ Less
Submitted 29 May, 2014;
originally announced May 2014.
-
CIWS-FW: a Customizable InstrumentWorkstation Software Framework for instrument-independent data handling
Authors:
Vito Conforti,
Massimo Trifoglio,
Andrea Bulgarelli,
Fulvio Gianotti,
Enrico Franceschi,
Luciano Nicastro,
Andrea Zoli,
Mauro Dadina,
Ricky Smart,
Roberto Morbidelli,
Marco Frailis,
Stefano Sartor,
Andrea Zacchei,
Marcello Lodi,
Roberto Cirami,
Fabio Pasian
Abstract:
The CIWS-FW is aimed at providing a common and standard solution for the storage, processing and quick look at the data acquired from scientific instruments for astrophysics. The target system is the instrument workstation either in the context of the Electrical Ground Support Equipment for space-borne experiments, or in the context of the data acquisition system for instrumentation. The CIWS-FW c…
▽ More
The CIWS-FW is aimed at providing a common and standard solution for the storage, processing and quick look at the data acquired from scientific instruments for astrophysics. The target system is the instrument workstation either in the context of the Electrical Ground Support Equipment for space-borne experiments, or in the context of the data acquisition system for instrumentation. The CIWS-FW core includes software developed by team members for previous experiments and provides new components and tools that improve the software reusability, configurability and extensibility attributes. The CIWS-FW mainly consists of two packages: the data processing system and the data access system. The former provides the software components and libraries to support the data acquisition, transformation, display and storage in near real time of either a data packet stream and/or a sequence of data files generated by the instrument. The latter is a meta-data and data management system, providing a reusable solution for the archiving and retrieval of the acquired data. A built-in operator GUI allows to control and configure the IW. In addition, the framework provides mechanisms for system error and logging handling. A web portal provides the access to the CIWS-FW documentation, software repository and bug tracking tools for CIWS-FW developers. We will describe the CIWS-FW architecture and summarize the project status.
△ Less
Submitted 28 May, 2014;
originally announced May 2014.
