The NVIDIA PilotNet Experiments
Authors:
Mariusz Bojarski,
Chenyi Chen,
Joyjit Daw,
Alperen Değirmenci,
Joya Deri,
Bernhard Firner,
Beat Flepp,
Sachin Gogri,
Jesse Hong,
Lawrence Jackel,
Zhenhua Jia,
BJ Lee,
Bo Liu,
Fei Liu,
Urs Muller,
Samuel Payne,
Nischal Kota Nagendra Prasad,
Artem Provodin,
John Roach,
Timur Rvachov,
Neha Tadimeti,
Jesper van Engelen,
Haiguang Wen,
Eric Yang,
Zongyi Yang
Abstract:
Four years ago, an experimental system known as PilotNet became the first NVIDIA system to steer an autonomous car along a roadway. This system represents a departure from the classical approach for self-driving in which the process is manually decomposed into a series of modules, each performing a different task. In PilotNet, on the other hand, a single deep neural network (DNN) takes pixels as i…
▽ More
Four years ago, an experimental system known as PilotNet became the first NVIDIA system to steer an autonomous car along a roadway. This system represents a departure from the classical approach for self-driving in which the process is manually decomposed into a series of modules, each performing a different task. In PilotNet, on the other hand, a single deep neural network (DNN) takes pixels as input and produces a desired vehicle trajectory as output; there are no distinct internal modules connected by human-designed interfaces. We believe that handcrafted interfaces ultimately limit performance by restricting information flow through the system and that a learned approach, in combination with other artificial intelligence systems that add redundancy, will lead to better overall performing systems. We continue to conduct research toward that goal.
This document describes the PilotNet lane-keeping effort, carried out over the past five years by our NVIDIA PilotNet group in Holmdel, New Jersey. Here we present a snapshot of system status in mid-2020 and highlight some of the work done by the PilotNet group.
△ Less
Submitted 17 October, 2020;
originally announced October 2020.
A submillimetre-bright z~3 overdensity behind a z~1 supercluster revealed by SCUBA-2 and Herschel
Authors:
A. G. Noble,
J. E. Geach,
A. J. van Engelen,
T. M. A. Webb,
K. E. K. Coppin,
A. Delahaye,
D. G. Gilbank,
M. D. Gladders,
R. J. Ivison,
Y. Omori,
H. K. C. Yee
Abstract:
We present a wide-field (30' diameter) 850um SCUBA-2 map of the spectacular three-component merging supercluster, RCS 231953+00, at z=0.9. The brightest submillimetre galaxy (SMG) in the field (S_850=12mJy) is within 30" of one of the cluster cores (RCS 2319-C), and is likely to be a more distant, lensed galaxy. Interestingly, the wider field around RCS 2319-C reveals a local overdensity of SMGs,…
▽ More
We present a wide-field (30' diameter) 850um SCUBA-2 map of the spectacular three-component merging supercluster, RCS 231953+00, at z=0.9. The brightest submillimetre galaxy (SMG) in the field (S_850=12mJy) is within 30" of one of the cluster cores (RCS 2319-C), and is likely to be a more distant, lensed galaxy. Interestingly, the wider field around RCS 2319-C reveals a local overdensity of SMGs, exceeding the average source density by a factor of 4.5, with a <1 per cent chance of being found in a random field. Utilizing Herschel-SPIRE observations, we find three of these SMGs have similar submillimetre colours. We fit their observed 250-850um spectral energy distributions to estimate their redshift, yielding 2.5<z<3.5, and calculate prodigious star formation rates (SFRs) ranging from 500-2500 solar masses per year. We speculate that these galaxies are either lensed SMGs, or signpost a physical structure at z~3: a 'protocluster' inhabited by young galaxies in a rapid phase of growth, destined to form the core of a massive galaxy cluster by z=0.
△ Less
Submitted 5 August, 2013;
originally announced August 2013.
