Results 1  10
of
13
Approximate Representations for MultiRobot Control Policies that Maximize Mutual Information
"... Abstract—We address the problem of controlling a small team of robots to estimate the location of a mobile target using nonlinear rangeonly sensors. Our control law maximizes the mutual information between the team’s estimate and future measurements over a finite time horizon. Because the computat ..."
Abstract

Cited by 13 (3 self)
 Add to MetaCart
(Show Context)
Abstract—We address the problem of controlling a small team of robots to estimate the location of a mobile target using nonlinear rangeonly sensors. Our control law maximizes the mutual information between the team’s estimate and future measurements over a finite time horizon. Because the computations associated with such policies scale poorly with the number of robots, the time horizon associated with the policy, and typical nonparametric representations of the belief, we design approximate representations that enable realtime operation. The main contributions of this paper include the control policy, an algorithm for approximating the belief state with provable error bounds, and an extensive study of the performance of these algorithms using simulations and real world experiments in complex, indoor environments. I.
A decentralized control policy for adaptive information gathering in hazardous environments
 in 51st IEEE Conf. Decision and Control, 2012
"... Abstract — This paper proposes an algorithm for driving a group of resourceconstrained robots with noisy sensors to localize an unknown number of targets in an environment, while avoiding hazards at unknown positions that cause the robots to fail. The algorithm is based upon the analytic gradient o ..."
Abstract

Cited by 7 (4 self)
 Add to MetaCart
(Show Context)
Abstract — This paper proposes an algorithm for driving a group of resourceconstrained robots with noisy sensors to localize an unknown number of targets in an environment, while avoiding hazards at unknown positions that cause the robots to fail. The algorithm is based upon the analytic gradient of mutual information of the target locations and measurements and offers two primary improvements over previous algorithms [5], [12]. Firstly, it is decentralized. This follows from an approximation to mutual information based upon the fact that the robots ’ sensors and environmental hazards have a finite area of influence. Secondly, it allows targets to be localized arbitrarily precisely with limited computational resources. This is done using an adaptive cellular decomposition of the environment, so that only areas that likely contain a target are given finer resolution. The estimation is built upon finite set statistics, which provides a rigorous, probabilistic framework for multitarget tracking. The algorithm is shown to perform favorably compared to existing approximation methods in simulation. I.
Nonparametric inference and coordination for distributed robotics
 In Proceedings of the IEEE Conference on Decision and Control
, 2012
"... Abstract — This paper presents nonparametric methods to infer the state of an environment by distributively controlling robots equipped with sensors. Each robot represents its belief of the environment state with a weighted sample set, which is used to draw likely observations to approximate the gr ..."
Abstract

Cited by 4 (2 self)
 Add to MetaCart
(Show Context)
Abstract — This paper presents nonparametric methods to infer the state of an environment by distributively controlling robots equipped with sensors. Each robot represents its belief of the environment state with a weighted sample set, which is used to draw likely observations to approximate the gradient of mutual information. The gradient leads to a novel distributed controller that continuously moves the robots to maximize the informativeness of the next joint observation, which is then used to update the weighted sample set via a sequential Bayesian filter. The incorporated nonparametric methods are able to robustly represent the environment state and robots’ observations even when they are modeled as continuousvalued random variables having complicated multimodal distributions. In addition, a consensusbased algorithm allows for the distributed approximation of the joint measurement probabilities, where these approximations provably converge to the true probabilities even when the number of robots, the maximum in/out degree, and the network diameter are unknown. The approach is implemented for five quadrotor flying robots deployed over a large outdoor environment, and the results of two separate exploration tasks are discussed. I.
Distributed Approximation of Joint Measurement Distributions Using Mixtures of Gaussians
"... Abstract—This paper presents algorithms to distributively approximate the continuous probability distribution that describes the fusion of sensor measurements from many networked robots. Each robot forms a weighted mixture of scaled Gaussians to represent the continuous measurement distribution (i.e ..."
Abstract

Cited by 1 (1 self)
 Add to MetaCart
(Show Context)
Abstract—This paper presents algorithms to distributively approximate the continuous probability distribution that describes the fusion of sensor measurements from many networked robots. Each robot forms a weighted mixture of scaled Gaussians to represent the continuous measurement distribution (i.e., likelihood) of its local observation. From this mixture set, each robot then draws samples of Gaussian elements to enable the use of a consensusbased algorithm that evolves the corresponding canonical parameters. We show that these evolved parameters form a distribution that converges weakly to the joint of all the robots’ unweighted mixture distributions, which itself converges weakly to the joint measurement distribution as more system resources are allocated. The innovation of this work is the combination of samplebased sensor fusion with the notion of preconvergence termination without the risk of ‘doublecounting ’ any single observation. We also derive bounds and convergence rates for the approximated joint measurement distribution, specifically the elements of its information vectors and the eigenvalues of its information matrices. Most importantly, these performance guarantees do not come at a significant cost of complexity, since computational and communication complexity of the canonical parameters scales quadratically with respect to the Gaussian dimension, linearly with respect to the number of samples, and constant with respect to the number of robots. Results from numerical simulations for object localization are discussed using both Gaussians and mixtures of Gaussians. estimation calculations in a centralized manner, and then globally broadcast the results to enable the robots to better position their sensors. For large systems, the central processor quickly becomes a computational and communication bottleneck, and thus is not considered to be scalable [4]. ˆj dimension
A COMPARATIVE STUDY OF UNDERWATER ROBOT PATH PLANNING ALGORITHMS FOR ADAPTIVE SAMPLING IN A NETWORK OF SENSORS
, 2014
"... Monitoring lakes, rivers, and oceans is critical to improving our understanding of complex largescale ecosystems. We introduce a method of underwater monitoring using semimobile underwater sensor networks and mobile underwater robots in this thesis. The underwater robots can move freely in all dim ..."
Abstract

Cited by 1 (1 self)
 Add to MetaCart
(Show Context)
Monitoring lakes, rivers, and oceans is critical to improving our understanding of complex largescale ecosystems. We introduce a method of underwater monitoring using semimobile underwater sensor networks and mobile underwater robots in this thesis. The underwater robots can move freely in all dimension while the sensor nodes are anchored to the bottom of the water column and can move only up and down along the depth of the water column. We develop three different algorithms to optimize the path of the
Adaptive Information Gathering Using Visual Sensors
"... Abstract — This paper proposes an algorithm to drive a robot equipped with a noisy, visual sensor to localize an unknown number of objects in an environment. The control strategy is based upon the analytic gradient of mutual information between the sensor readings and the object locations. An adapti ..."
Abstract
 Add to MetaCart
(Show Context)
Abstract — This paper proposes an algorithm to drive a robot equipped with a noisy, visual sensor to localize an unknown number of objects in an environment. The control strategy is based upon the analytic gradient of mutual information between the sensor readings and the object locations. An adaptive cellular decomposition is used to represent the environment, increasing resolution only in regions likely to contain an object. The unknown number and locations of both objects and sensor readings are modeled using random finite sets and a recursive Bayesian filter maintains the robot’s belief over the distribution of object locations. Utilizing the fact that a visual sensor can only see a finite subset of the whole environment, the complexity of the Bayesian filter update and mutual information gradient computations are significantly reduced. Numerical simulations and experimental results are used to illustrate the performance of the filter and controller. I.
Towards a Unifying Information Theoretic Framework for MultiRobot Exploration and
"... In this talk we discuss our recent work on a mathematical framework for pursuing exploration and surveillance tasks using multiple collaborating robots. We ground this framework in the first principles of information theory, and in doing so establish a unifying model that considers the interdepende ..."
Abstract
 Add to MetaCart
(Show Context)
In this talk we discuss our recent work on a mathematical framework for pursuing exploration and surveillance tasks using multiple collaborating robots. We ground this framework in the first principles of information theory, and in doing so establish a unifying model that considers the interdependencies of system resources pertaining to robot mobility, sensing, and communication. The framework identifies metrics that characterize system performance and provides qualitative understanding of quantitative results. We show that exploration and surveillance1 can be considered close relatives who can both be described with the same framework, and as a result approaches
Pipelined Consensus for Global State Estimation in MultiAgent Systems
"... This paper presents pipelined consensus, a practical, robust consensus algorithm for multiagent systems using mesh networks. During each round, each agent starts a new consensus. Each agent maintains the intermediate results for the previous k consensus in a pipeline message. After k rounds, the ..."
Abstract
 Add to MetaCart
(Show Context)
This paper presents pipelined consensus, a practical, robust consensus algorithm for multiagent systems using mesh networks. During each round, each agent starts a new consensus. Each agent maintains the intermediate results for the previous k consensus in a pipeline message. After k rounds, the results of the first consensus are ready. The pipeline keeps each consensus independent, so any errors only persist for k rounds. This makes pipelined consensus robust to many realworld problems that other algorithms cannot handle, including message loss, changes in network topology, sensor variance, and changes in agent population. The algorithm is fully distributed and selfstabilizing, and uses a communication message of fixed size. We demonstrate the efficiency of pipelined consensus in two scenarios: computing mean sensor values in a distributed sensor network, and computing a centroid estimate in a multirobot system. We provide extensive simulation results, and realworld experiments with up to 24 agents. The algorithm produces accurate results, and it handles all of the disturbances mentioned above.
46 10709932/14/$31.00©2014IEEEt IEEE ROBOTICS & AUTOMATION MAGAZINE t JUNE 2014 By
"... T he task addressed in this article is the localization of an unknown number of targets using a mobile robot equipped with a visual sensor. The estimation of the number of targets and their locations is done using a recursive Bayesian filter over random finite sets (RFSs), and the position of the ro ..."
Abstract
 Add to MetaCart
(Show Context)
T he task addressed in this article is the localization of an unknown number of targets using a mobile robot equipped with a visual sensor. The estimation of the number of targets and their locations is done using a recursive Bayesian filter over random finite sets (RFSs), and the position of the robot is assumed to be known. We present a computationally tractable control law whereby the robot follows the gradient of mutual information between target locations and detections. The method is verified through realworld experimental trials, reliably detecting multiple targets and ignoring clutter obstacles. A Game of Fetch The ability for robots to locate and interact with objects of interest within an unstructured environment is very important as robots move out of controlled settings. In this article, we examine a prototypical example of playing fetch with a robot. First the robot is shown a new object and then it must go into the field and locate a small yet potentially unknown number of these objects that are scattered throughout the environment. After locating all of the objects, the robot collects them and returns them to the user. Such behavior has obvious extensions to household robots, inspection tasks, and search and rescue. Using realworld experiments with the robot shown in Figure 1, we present results showing the localization of a variety of objects, focusing on the control and estimation rather than the collection and return tasks. The use of Bayesian filtering to estimate unknown and uncertain environments is well established, with many current methods summarized by Thrun et al. [19]. In particular, the problem of multiobject tracking has been addressed in several contexts, including simultaneous localization and mapping (SLAM), computer vision, and radarbased tracking, using a