Matthew Ridley

The ANSER Project: Data Fusion Across Multiple Uninhabited Air Vehicles

The objective of the autonomous navigation and sensing experiment research (ANSER) project is to demonstrate decentralized data fusion (DDF) and simultaneous localization and map building (SLAM) across multiple uninhabited air vehicles (UAVs). To achieve this objective, the project specifies the development of four UAVs, where each UAV houses up to two terrain sensors and an INS/GPS navigation system. The terrain sensors include a scanning radar, laser/vision and standard vision system. The DDF concept has to be shown to be effective both on a single UAV and on multiple UAVs. The proof of the concept will lie in the ability of the DDF structure to conduct multi-target tracking problems as well as SLAM. To obtain this goal, a number of subgoals are required, most of which have never been attempted before on a research level. The objective of this paper is to present these goals as an overview of the ANSER project along with some simulated and real-time results.

Consistent methods for Decentralised Data Fusion using Particle Filters

This paper presents two solutions for performing decentralised particle filtering in view of non-linear, non-Gaussian tracking in sensor networks. The issue is that no known methods exist to deal with correlated estimation errors due to common past information between two discrete particle sets. The first method transforms the particles to a Gaussian mixture model, the second approximates the set by a Parzen density estimate. Both of these representations accommodate consistent fusion and maintain accurate summaries of the particles. Requiring less bandwidth than particle representations, transformations to GMMs or Parzen representations for communication provide an added advantage. The accuracy in which the algorithms summarise the particle set, fusion methods and bandwidth requirements of each representation will be compared. Our results show that whilst less GMM components are required to summarise the sample statistics, the decentralised fusion solution using Parzen representations yields a more accurate result

Rich probabilistic representations for bearing only decentralised data fusion

The aim of this paper is to demonstrate the validity of using Gaussian mixture models (GMM) for representing probabilistic distributions in a decentralised data fusion (DDF) framework. GMMs are a powerful and compact stochastic representation allowing efficient communication of feature properties in large scale decentralised sensor networks. It will be shown that GMMs provide a basis for analytical solutions to the update and prediction operations for general Bayesian filtering. Furthermore, a variant on the covariance intersect algorithm for Gaussian mixtures will be presented ensuring a conservative update for the fusion of correlated information between two nodes in the network. In addition, purely visual sensory data will be used to show that decentralised data fusion and tracking of non-Gaussian states observed by multiple autonomous vehicles is feasible.

Tracking in decentralised air-ground sensing networks

This paper describes the theoretical and practical development of a decentralised air and ground sensing network for target tracking and identification. The theoretical methods employed for studying decentralised data fusion problems are based on the information-filter formulation of the Kalman filter algorithm and on information-theoretic methods derived from the Bayes theorem. The paper particularly focuses on how these methods are applied in very large heterogeneous sensor networks, where there may be a significant amount of data delay or corruption in communication. This paper then describes the development of a practical system aimed at demonstrating some of these principles. The system consists of a number of unmanned air vehicles (UAVs), with radar and vision payloads, able to observe a number of ground targets. The UAV sensor payloads are constructed in a modular fashion, with the ability to communicate in a network with both other air-borne and other ground sensors. The ground sensor system comprises of multiple modular sensing nodes which include vision scanned laser, steerable radar, multiple fixed radar arrays, and combined night vision (IR)-radar.

A decentralised particle filtering algorithm for multi-target tracking across multiple flight vehicles

This paper presents a decentralised particle filtering algorithm that enables multiple vehicles to jointly track 3D features under limited communication bandwidth. This algorithm, applied within a decentralised data fusion (DDF) framework, deals with correlated estimation errors due to common past information when fusing two discrete particle sets. Our solution is to transform the particles into Gaussian mixture models (GMMs) for communication and fusion. Not only can decentralised fusion be approximated by GMMs, but this representation also provides summaries of the particle set. Less bandwidth per communication step is required to communicate a GMM than the particle set itself hence conversion to GMMs for communication is an advantage. Real airborne data is used to demonstrate the accuracy of our decentralised particle filtering algorithm for airborne tracking and mapping

Real time Multi-UAV Simulator

This paper presents the system architecture of a real time multi-UAV simulator (RMUS). The simulator has been implemented as both a testing and validation mechanism for the real demonstration of multiple UAVs conducting both decentralised data fusion and control. These mechanisms include the off-line simulation of complex scenarios, hardware-in-the-loop tests, validation of real test results, and online mission control system demonstrations. The paper also present CommLibX, a novel communication framework for the system which allows simulation modules to communicate over single or multiple virtual channels. This unique communication system is then easily ported onto the real hardware allowing for maximum reuse of software and integrity.

Decentralised ground target tracking with heterogeneous sensing nodes on multiple UAVs

This paper presents real time results of a decentralised air-borne data fusion system tracking multiple ground based targets. These target estimates are then used to construct a map of the environment. A decentralised communication strategy is employed which is robust to communication latencies and dropouts and results in each sensing node having a local estimate using global information. In addition, this paper describes both hardware and algorithms used to deploy two sensor nodes for such a task. Two sensor types will be discussed, vision and mm wave radar. The problems introduced by locating the sensors on air vehicles are both interesting and challenging. A total of four unmanned air vehicles will be employed to carry node payloads. Weight and power restrictions of the payloads coupled with the vehicle dynamics make the task of processing and fusing vision and radar based data a challenging problem indeed. This paper aims to highlight many of the problems that have been encountered in developing both hardware and software to operate under such constraints.

A Comparison of Probabilistic Representations for Decentralised Data Fusion

This paper compares and constrasts three different probabilistic models - Particle representations, Parzen density estimates, and Gaussian mixture models - for non-Gaussian, non-linear feature tracking, when applied to multiple autonomous vehicles using the Decentralised Data Fusion (DDF) paradigm. These probabilistic models were chosen as they are all capable of approximating the probability distributions of an ideal Bayesian filter and have different properties with regard to computational efficiency and quality of the approximation. In order to show that each model satisfy the DDF requirements of modularity, scalability and robustness, the performance of each representation is taken from a simulation for multi-sensor bearing-only tracking. Performance is evaluated in three areas: (a) mathematical accuracy and optimality of fusion for correlated information between nodes, (b) computational efficiency and accuracy of various operations in the DDF framework and (c) bandwidth requirements for communicating the representations over a wireless network.

Decentralised data fusion with Parzen density estimates

Decentralised sensor networks typically consist of multiple processing nodes supporting one or more sensors. These nodes are interconnected via wireless communication. Practical applications of decentralised data fusion have generally been restricted to using Gaussian based approaches such as the Kalman or information filter. This paper proposes the use of Parzen window estimates as an alternate representation to perform decentralised data fusion. It is required that the common information between two nodes be removed from any received estimates before local data fusion may occur. Otherwise, estimates may become overconfident due to data incest. A closed form approximation to the division of two estimates is described to enable conservative assimilation of incoming information to a node in a decentralised data fusion network. A simple example of tracking a moving particle with Parzen density estimates is shown to demonstrate how this algorithm allows conservative assimilation of network information.

Multi-level State Estimation in an Outdoor Decentralised Sensor Network

Decentralised estimation of heterogeneous sensors is performed on an outdoor network. Attributes such as position, appearance, and identity represented by non-Gaussian distributions are used in in the fusion process. It is shown here that real-time decentralised data fusion of non-Gaussian estimates can be used to build rich environmental maps. Human operators are also used as additional sensors in the network to complement robotic information.