Self-Organising Zooms for Decentralised Redundancy Management in Visual Sensor Networks

Abstract

Abstract-When visual sensor networks are composed of cameras which can adjust the zoom factor of their own lens, one must determine the optimal zoom levels for the cameras, for a given task. This gives rise to an important trade-off between the overlap of the different cameras' fields of view, providing redundancy, and image quality. In an object tracking task, having multiple cameras observe the same area allows for quicker recovery, when a camera fails. In contrast having narrow zooms allow for a higher pixel count on regions of interest, leading to increased tracking confidence. In this paper we propose an approach for the self-organisation of redundancy in a distributed visual sensor network, based on decentralised multi-objective online learning using only local information to approximate the global state. We explore the impact of different zoom levels on these trade-offs, when tasking omnidirectional cameras, having perfect 360-degree view, with keeping track of a varying number of moving objects. We further show how employing decentralised reinforcement learning enables zoom configurations to be achieved dynamically at runtime according to an operator's preference for maximising either the proportion of objects tracked, confidence associated with tracking, or redundancy in expectation of camera failure. We show that explicitly taking account of the level of overlap, even based only on local knowledge, improves resilience when cameras fail. Our results illustrate the trade-off between maintaining high confidence and object coverage, and maintaining redundancy, in anticipation of future failure. Our approach provides a fully tunable decentralised method for the self-organisation of redundancy in a changing environment, according to an operator's preferences.

On Learning in Collective Self-Adaptive Systems: State of Practice and a 3D Framework

“…inclusion criterion IC-4 in Table II). We found only six studies (∼12% in Table III) explicitly modelling CSAS agents with restricted autonomy [34], [38], [45], [58], [66], [72]. Within a restricted autonomy setting, there is at least one agent whose actions are supervised and could be overwritten by other agents.…”

“…Another group of studies [28], [33], [38], [49], [53], [73], [77] consider CSAS in which agents share limited information and not their full state. This limited information could take the form of selected agent configurations [73], adaptation tactics [28] or expected rewards [33], [38], [53]. Although the term can subsume many degrees of knowledge access, information sharing between agents in these studies is not enough to directly influence the adopted learning processes.…”

Centralised, Decentralised, and Self-Organised Coverage Maximisation in Smart Camera Networks

“…The W-Learner exchanges information about current states with neighbouring cameras to learn the optimal placement over time. In [13], we investigated the self-organised coverage in 360…”

The Future of Camera Networks

“…However, when networked, they can cooperate in order to carry out more complex tasks and achieve collective goals [13,14,27,36,46]. As with other types of cyber-physical system, more recently, self-awareness [24,29] has been introduced in smart cameras, enabling them to reason about their own behaviour, and adapt accordingly in changing environments [15,42].…”

“…Barrier optimization refines this to ensure objects cannot pass between areas undetected. k refers to the number of cameras observing the same area at any given time [8,11,14].…”