We present an approach to capture the 3D motion of a group of people engaged in a social interaction, where inter-occlusions are frequent and functional. The Panoptic Studio is a system organized around the thesis that social interactions should be measured through the integration of perceptual analyses over a large variety of viewpoints. We present a modularized system designed around this principle, consisting of integrated structural, hardware, and software innovations. The system takes, as input, 480 synchronized video streams of multiple people engaged in social activities, and produces, as output, the labeled time-varying 3D structure of anatomical landmarks on individuals in the space. Our algorithm is designed to fuse the "weak" perceptual processes in the large number of views by progressively generating skeletal proposals from low-level appearance cues, and a framework for temporal refinement is also presented by associating body parts to reconstructed dense 3D trajectory stream. Our system and method are the first in reconstructing full body motion of more than five people engaged in social interactions without using markers. We also empirically demonstrate the impact of the number of views in achieving this goal.
In the extreme environments posed by war fighting, fire fighting, and nuclear accident response, the cost of direct human exposure is levied in terms of injury and death. Robotic alternatives must address effective operations while removing humans from danger. This is profoundly challenging, as extreme environments inflict cumulative performance damage on exposed robotic agents. Sensing and perception are among the most vulnerable components. We present a distributed robotic system that enables autonomous reconnaissance and mapping in urban structures using teams of robots. Robot teams scout remote sites, maintain operational tempos, and successfully execute tasks, principally the construction of 3-D Maps, despite multiple agent failures. Using an economic model of agent interaction based on a free market architecture, a virtual platform (a robot colony) is synthesized where task execution does not directly depend on individual agents within the colony. MOTIVATIONMilitary Operations in Urban Terrain (MOUT) pose fierce constraints such as limited visibility, complex and expansive fortifications, limited intelligence, and the presence of native populations and other non-combatants that prohibit deployment of large forces [1,2]. Further, the use of asymmetric threats, e.g. biological and chemical agents, against both land forces and indigenous populations in urban settings is an increasing likelihood [3]. These conditions place land forces and noncombatants in a highly non-deterministic, dangerous, confrontational, and volatile environment. An effort to identify and improve the ability of ground forces to project sufficient force and safeguard non-combatants is underway. This program, called the MOUT Advanced Concept Technology Development (ACTD), focuses on improving operational effectiveness in urban areas [4,5].The development of robotics technology will enable minimally invasive and precise MOUT operations that reduce risk to both ground forces and non-combatants by removing soldiers from dangerous and sometimes confrontational tasks [6]. Potential tasks for robotic systems include mine sweeping, reconnaissance, security and monitoring presence, and communications infrastructure [7]. DARPA has undertaken the task of enabling distributed robotics technology to execute urban reconnaissance missions. As a part of this effort, DARPA's Software for Distributed Robotics (SDR) program is looking at revolutionary approaches to the development of multi-agent robotic systems within this task domain. Under the auspices of the SDR Program, our team is producing software technology designed to construct interior maps of urban structures using groups of homogenous mobile robots. This concept software will be demonstrated on surrogate, simplistic mobile robotic platforms and eventually ported to mission capable mobile robotic systems.Our approach to the problem focuses initially on the production of reliable, accurate, and fault-tolerant software by exploiting the redundancy inherent in a group of homogenous robots, ...
The "Five Point Relative Pose Problem" is to find all possible camera configurations between two calibrated views of a scene given five point-correspondences. We take a fresh look at this well-studied problem with an emphasis on the parametrization of Essential Matrices used by various methods over the years. Using one of these parametrizations, a novel algorithm is proposed, in which the solution to the problem is encoded in a system of nine quadratic equations in six variables, and is reached by formulating this as a constrained optimization problem. We compare our algorithm with an existing 5-point method, and show our formulation to be more robust in the presence of noise.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.