Distributed trajectory estimation with privacy and communication constraints: A two-stage distributed Gauss-Seidel approach

Choudhary, Siddharth; Carlone, Luca; Nieto, Carlos; Rogers, John G.; Christensen, Henrik I.; Dellaert, Frank

doi:10.1109/icra.2016.7487736

Cited by 30 publications

(25 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The relative pose estimation module (RelPose) extracts the parts of the sparse map that are observed at the candidate place matches and uses them to establish relative poses between the robots trajectories, or to reject candidate matches. The decentralized optimization module (DOpt) [2] obtains the initial guess from the map, intra-robot relative pose measurements E I from the VO and the inter-robot relative pose measurements E S from RelPose, and updates the map. The system works continuously as new images are acquired.…”

Section: Methodsmentioning

confidence: 99%

“…Several systems have previously combined decentralized optimization and relative pose estimation into decentralized SLAM. Several optimization works cited in Section II-A, such as [2], [4] use direct relative measurements and can thus be considered full decentralized SLAM systems with the recall limitation illustrated in Fig. 2.…”

Section: Integrated Decentralized Slammentioning

confidence: 99%

“…Visual odometry is not involved in the feedback loop and is thus trivial to distribute -it can be run on each robot independently and feed its output to the rest of the decentralized SLAM system. In previous work we have proposed state-of-the art decentralized place recognition [1] and optimization [2] systems separately. Both systems focus on data efficiency: they achieve performance similar to a centralized system while minimizing the data that needs to be exchanged.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Data-Efficient Decentralized Visual SLAM

Cieslewski

Choudhary

Scaramuzza

2018

2018 IEEE International Conference on Robotics and Automation (ICRA)

Self Cite

172

150

View full text Add to dashboard Cite

Decentralized visual simultaneous localization and mapping (SLAM) is a powerful tool for multi-robot applications in environments where absolute positioning systems are not available. Being visual, it relies on cameras, cheap, lightweight and versatile sensors, and being decentralized, it does not rely on communication to a central ground station. In this work, we integrate state-of-the-art decentralized SLAM components into a new, complete decentralized visual SLAM system. To allow for data association and co-optimization, existing decentralized visual SLAM systems regularly exchange the full map data between all robots, incurring large data transfers at a complexity that scales quadratically with the robot count. In contrast, our method performs efficient data association in two stages: in the first stage a compact full-image descriptor is deterministically sent to only one robot. In the second stage, which is only executed if the first stage succeeded, the data required for relative pose estimation is sent, again to only one robot. Thus, data association scales linearly with the robot count and uses highly compact place representations. For optimization, a state-of-the-art decentralized pose-graph optimization method is used. It exchanges a minimum amount of data which is linear with trajectory overlap. We characterize the resulting system and identify bottlenecks in its components. The system is evaluated on publically available data and we provide open access to the code.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Integrated Decentralized Slammentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Data-Efficient Decentralized Visual SLAM

Cieslewski

Choudhary

Scaramuzza

2018

2018 IEEE International Conference on Robotics and Automation (ICRA)

Self Cite

172

150

View full text Add to dashboard Cite

show abstract

“…The second reason is that Gaussian elimination is performed on a linearized version of the problem, hence approaches such as DDF-SAM [54] require good linearization points and complex bookkeeping to ensure consistency of the linearization points across the robots. An alternative approach to Gaussian elimination is the Gauss-Seidel approach of Choudhary et al [47], which implies a communication burden which is linear in the number of separators.…”

Section: Long-term Autonomy Ii: Scalabilitymentioning

confidence: 99%

Past, Present, and Future of Simultaneous Localization and Mapping: Toward the Robust-Perception Age

Cadena¹,

Carlone²,

Carrillo³

et al. 2016

IEEE Trans. Robot.

Self Cite

2,789

1,742

View full text Add to dashboard Cite

Abstract-Simultaneous localization and mapping (SLAM) consists in the concurrent construction of a model of the environment (the map), and the estimation of the state of the robot moving within it. The SLAM community has made astonishing progress over the last 30 years, enabling large-scale real-world applications and witnessing a steady transition of this technology to industry. We survey the current state of SLAM and consider future directions. We start by presenting what is now the de-facto standard formulation for SLAM. We then review related work, covering a broad set of topics including robustness and scalability in long-term mapping, metric and semantic representations for mapping, theoretical performance guarantees, active SLAM and exploration, and other new frontiers. This paper simultaneously serves as a position paper and tutorial to those who are users of SLAM. By looking at the published research with a critical eye, we delineate open challenges and new research issues, that still deserve careful scientific investigation. The paper also contains the authors' take on two questions that often animate discussions during robotics conferences: Do robots need SLAM? and Is SLAM solved?

show abstract

“…Compute prediction message m fX i !Xi (X t i ) and m fL k !Lk (L t k ) by (15) and (20), respectively 3 Broadcast prediction distributions as initial messages of iteration process by (36)- (38) 4 for iteration r = 1 : R do 5…”

Section: Resultsmentioning

confidence: 99%

Distributed simultaneous localization and mapping for mobile robot networks via hybrid dynamic belief propagation

Wan

et al. 2017

International Journal of Distributed Sensor Networks

View full text Add to dashboard Cite

This article proposes a hybrid dynamic belief propagation for simultaneous localization and mapping in the mobile robot network. The positions of landmarks and the poses of moving robots at each time slot are estimated simultaneously in an online and distributed manner, by fusing the odometry data of each robot and the measurements of robot-robot or robot-landmark relative distance and angle. The joint belief state of all robots and landmarks is encoded by a factor graph and the marginal posterior probability distribution of each variable is inferred by belief propagation. We show how to calculate, broadcast, and update messages between neighboring nodes in the factor graph. Specifically, we combine parametric and nonparametric techniques to tackle the problem arisen from non-Gaussian distributions and nonlinear models. Simulation and experimental results on publicly available dataset show the validity of our algorithm.

show abstract

Distributed trajectory estimation with privacy and communication constraints: A two-stage distributed Gauss-Seidel approach

Cited by 30 publications

References 34 publications

Data-Efficient Decentralized Visual SLAM

Data-Efficient Decentralized Visual SLAM

Past, Present, and Future of Simultaneous Localization and Mapping: Toward the Robust-Perception Age

Distributed simultaneous localization and mapping for mobile robot networks via hybrid dynamic belief propagation

Contact Info

Product

Resources

About