Decentralised Solutions to the Cooperative Multi-Platform Navigation Problem

Abstract

The problem of cooperative navigation for a team of platforms employing inter-platform observations is investigated. A decentralised solution in the framework of an information filter with delayed states is presented. In this structure, each platform first estimates its motion using only local sensor data, then shares its information across the network using an algorithm that employs a distributed Cholesky modification. The decentralised solution permits each platform to act in the same modular manner, providing robustness to individual platform failure. The solution yields linear minimum mean-square error estimation performance. As such the estimates generated are optimal ; it generates exactly the same estimates as would a conventional extended Kalman filter, if given the same data. Efficient sparse implementation is accomplished without resorting to approximate methods. Simulation experiments employing a team of ten mobile platforms are described and used to evaluate the decentralised estimation performance. The robustness, flexibility and cost of the decentralised approach are analyzed and compared to an existing distributed solution.

Distributed collaborative localization for a heterogeneous multi-robot system

“…Although [11] demonstrated constant computational cost O(1), they neglect the possible interdependencies between robots' pose estimation leading for over confident state estimation. The computational complexity of the distributed CL approach presented in [12] increases with number of IRRM as they maintain the history of IRRM update.…”

Decentralised cooperative localisation for heterogeneous teams of mobile robots

“…A version of information-form cooperative localisation is presented in [13], where sparse Cholesky factors of the joint state estimate are pipelined from platform to platform. This approach has higher lag and computational expense than the current work, but in the current work we apply the idea of incremental Cholesky operations for efficient fusion and moment recovery.…”

“…In practice, we implement reconstruction and fusion using the incremental sparse Cholesky operations described in [13,Section 3.4]. The Cholesky-form shares the sparse additive properties on the information-form, but also permits efficient sparse recovery of mean and covariance components.…”

Cooperative navigation and coverage identification with random gossip and sensor fusion

“…Modern autonomous systems are deployed in various environments, such as underwater [1], indoors [2], or in civil and military aviation [3]. Recently, there has been an increased interest in multi-vehicle missions, where each platform acts autonomously to fulfill a specified task.…”

“…For instance, the same external landmarks may be used by different platforms for self-localization, or inter-platform observations are available. Exploiting this type of information generally improves the self-localization of each platform, and therefore navigation of multiple platforms need to be treated as a whole [2], which is called cooperative navigation.…”