Multi-Agent Data Collection in Non-Stationary Environments

Coordinated multi-robot systems are an effective way to harvest data from sensor networks and to implement active perception strategies. However, achieving efficient coordination in a way which guarantees a target QoS while adapting dynamically to changes (in the environment, due to sensors’ mobility, and/or in the value of harvested data) is to date a key open issue. In this paper, we propose a novel decentralized Monte Carlo Tree Search algorithm (MCTS) which allows agents to optimize their own actions while achieving some form of coordination, in a changing environment. Its key underlying idea is to balance in an adaptive manner the exploration-exploitation trade-off to deal effectively with abrupt changes caused by the environment and random changes caused by other agents’ actions. Critically, outdated and irrelevant samples - an inherent and prevalent feature in all multi-agent MCTS-based algorithms - are filtered out by means of a sliding window mechanism. We show both theoretically and through simulations that our algorithm provides a log-factor (in terms of time steps) smaller regret than state-of-the-art decentralized multi-agent planning methods. We instantiate our approach on the problem of underwater data collection, showing on a set of different models for changes that our approach greatly outperforms the best available algorithms for that setting, both in terms of convergence speed and of global utility.

I. Introduction

Over the last few years, multi-robot cooperative data harvesting, based on unmanned aerial vehicles (UAVs) or underwater autonomous vehicles (AUVs) has attracted widespread attention [1]. Autonomous vehicles (AVs) are widely used for efficient data communication as wireless relays. Thanks to their high mobility, they can move in proximity of IoT devices to harvest data and relay them to a fusion center. This allows avoiding multi-hop or long range transmissions, greatly prolonging IoT devices lifetime. In particular, decentralized collaborative multi-robot systems for active perception have become popular due to their robustness and scalability, which makes them suitable for industrial and military applications, as well as in environmental monitoring, in search and rescue missions, or in online object recognition and tracking [1]–[3]. The use of AVs is particularly beneficial when sensors/tracked objects are too far apart from each other and from data sinks, and when sensors location, tracked objects, and the environment around change over time in an unpredictable manner, such as in post-disaster (e.g. flood, earthquakes) scenarios. It is this variability over time which makes the problem of how to effectively achieve coordination in a decentralized manner – emerging from the need to guarantee a target performance, e.g. in terms of AoI or latency – a key issue still largely unsolved [4]–[6].