Report

Teaser, summary, work performed and final results

Periodic Reporting for period 1 - SOLOMON (Self-Organisation and Learning Online in Mobile Observation Networks)

Teaser

The SOLOMON project is driven by the rising demand for rapid-deployment camera networks which can adapt to provide security in the context of unforeseen situations and unfolding scenarios. This is evidenced by the rapid growth of leading body-cam company Edesix Ltd, whose...

Summary

The SOLOMON project is driven by the rising demand for rapid-deployment camera networks which can adapt to provide security in the context of unforeseen situations and unfolding scenarios. This is evidenced by the rapid growth of leading body-cam company Edesix Ltd, whose VideoBadge technology is being adopted by police forces worldwide. However, research advances in smart camera networks have not been realised in dynamic body-worn camera networks, and still rely on prohibitively expensive static hardware. In the SOLOMON project we aimed to develop a novel type of lightweight, inexpensive smart camera network suitable for rapid deployment and reconfiguration, where low-cost camera devices such as Edesix’s VideoBadge, are paired with the processing capabilities of smartphones. These are then worn by people (e.g. police, security guards) or mounted on mobile robots. This not only lowers cost, but allows us to introduce a feedback loop between the sensing cameras and the acting people/robots, enabling the camera network to adapt to changes during runtime, for example to prioritise or cover newly relevant areas, in response to rapidly unfolding situation. The main goal of the SOLOMON project was to develop novel techniques in collective decision making and self- organisation as well as multi-objective online learning, in order to achieve this vision.

Work performed

The work performed in the SOLOMON project shed light on several aspects of autonomous mobile smart cameras able to interact with their environment including other smart cameras and to deal with rapidly unfolding situations. Specifically, these aspects are (i) the ‘online multi-target k-coverage problem’, (ii) self-organisation to coordinate cameras in a network to achieve shared and individual goals, and (iii) self-awareness of computational systems embodied in physical devices.

First, at the beginning of the SOLOMON project, we proposed a completely novel problem statement, namely the ‘online multi-target k-coverage problem’ [1]. The paper introducing this problem and proposing these initial approaches has been nominated for the best paper award at the International Conference on Distributed Smart Cameras, the leading conference on smart camera networks.
We also proposed a novel architecture and investigated self-aware capabilities to be involved in mobile smart camera networks and more specifically in the online multi-object k-coverage problem [2]. In this work decisions on who to invite but also on how to respond to an invitation is based on local information. This increased the network-wide performance even further and showed the benefits and applicability of our novel architecture.

Second, we studied self-organisation in autonomous acting, mobile smart cameras.
We investigated the benefits of self-organising behaviour over decentralised and distributed behaviour in smart camera networks [3]. We took the problem of coverage maximisation and showed that self-organising, iterative techniques can even outperform concurrent, decentralised approaches with respect to coverage maximisation while minimising overlap with other cameras. This work was presented at the International Conference for Self-Adaptive and Self-Organising Systems, where the paper was nominated for the best paper award of the conference. Furthermore, we proposed a novel algorithm to coordinate cameras in a distributed fashion, guaranteeing their self-organisation to achieve the online multi-object k-coverage problem [4]. Finally, we investigated team performance of teams on the online multi-object k-coverage problem [5]. Our results showed the benefit of online affiliation over static or single teams.

Third, we investigated computational self-awareness in physical devices, interacting with the environment and with other systems. We defined levels of networked self-awareness describing the impact of actions in the environment on the system [6, 7]. Another article highlighted the challenges for cyber-physical systems and how computational self-awareness might help to overcome these challenges [8].

In total, the SOLOMON project produced 9 peer review papers accepted to 6 conferences and 3 workshops. These papers have been written together with partners from the SOLOMON project as well as collaborations established by the research fellow during the project runtime. Furthermore, there are 2 conference papers and 3 journal articles currently under review.

[1] “Online Multi-object k-coverage with Mobile Smart Cameras”, Lukas Esterle, Peter R. Lewis. In Proceedings of the International Conference on Distributed Smart Cameras, 2017
[2] “An Architecture for Self-Aware IoT Systems”, Lukas Esterle and Bernhard Rinner. In Proceedings of the International Conference on Acoustics, Speech and Signal Processing, 2018
[3] “Centralised, Decentralised, and Self-organised Coverage Maximisation in Smart Camera Networks”, Lukas Esterle. In Proceedings of the International Conference on Self-Adaptive and Self-Organised Systems, 2017
[4] “CHAINMAIL: Distributed coordination for multi-task k-assignment using autonomous mobile IoT devices”, Lukas Esterle. In Proceedings of the International Conference on Distributed Computation in Sensor Systems, 2018
[5] “Goal-aware Team Affiliation in Collectives of Autonomous Robots”, Lukas Esterle. In Proceedings

Final results

The SOLOMON project contributed to the progress beyond the state-of-the-art in several important aspects:

Our first contribution beyond the state-of-the-art was to defined a new problem formulation: online multi-object k-coverage problem. This problem brings together the k-coverage problem from sensor networks with the CMOMMT problem (cooperative multi-robot observation of multiple moving targets). Furthermore, we provided a novel coordination algorithm inspired by gossiping mechanisms to tackle this problem.
In our second major contribution, we showed the benefits of self-organising behaviour for the coverage maximisation problem.
Our third contribution explored the benefits of teams tackling multiple tasks concurrently to achieve an overall goal. We not only showed that multiple teams perform better than single teams but also that dynamic team formation, where agents choose and change their team during runtime, easily outperforms agents where the team was assigned at deployment time.
Finally, we proposed networked self-awareness and a corresponding architecture for autonomous agents to specifically consider their immediate environment.

During the project Edesix was bought by Motorola, sowing down the application of developed approaches. Nevertheless, the research performed will lead to quicker deployment of mobile camera networks yielding in lower deployment costs.

Website & more info

More info: https://alice.aston.ac.uk/solomon/.