Collecte et traitement des informations en fonction du contexte en vue de soutenir les systèmes autonomes dans les scénarios de l'industrie 4.0
Razanne Abu-Aisheh · HAL (Le Centre pour la Communication Scientifique Directe) · 2023
Industry 4.0 environments are characterized by the coexistence of a diverse set of devices, including sensors, mixed-reality displays, robots, drones, and smart objects. These systems must be capable of autonomously taking critical in-time decisions necessary to perform complex tasks without human input. One essential application for Industry 4.0 is multi-robot exploration and mapping of unknown environments, especially in critical missions such as hazard detection and search and rescue. These missions share the need to reach full coverage of the explorable space in the shortest time possible. To minimize completion time, robots in the fleet must be able to exchange information about the environment reliably with one another. However, existing exploration and mapping algorithms suffer from inaccuracies and inefficiencies due to their lack of contextual awareness of their surroundings, especially in terms of communications, lacking flexibility and adaptability to the environment, and hence, adding unnecessary delay to the mission at hand. In this thesis, we investigate the impact of communication awareness on the performance of multi-robot exploration and mapping expeditions, in terms of time to completion. We evaluate existing research in the field and demonstrate the impact of not considering communication impairments when designing such algorithms. From there, we propose Atlas, an exploration and mapping algorithm that natively takes packet loss into account, with a 100% completion ratio even with Packet Delivery Ratios (PDRs) as low as 0.1. However, Atlas on its own cannot handle scenarios where connectivity is completely lost. It also adds a significant delay to the completion of the mission, as lost packets keep getting re-transmitted periodically until they are received. One solution is relay placement. Most research on relay placement for multi-robot expeditions tend to fall into two categories. First, communication-aware relay placement based on initial Received Signal Strength Indicator (RSSI) is used. However, this requires running a full mission prior to the exploration to find the optimal position for the relays to be placed. Second, maintaining a distance (specified prior to the mission) between relays and exploration robots. These methods add to the time it takes to complete the mission. The research question becomes how can we place relays to maintain communication as reliable as possible, and also dynamically throughout the exploration mission without prior knowledge of the environment, in a way that reduces delay to the exploration and mapping time to completion. We solve this by proposing ``Connectivity Aware Relay Algorithm'' (CARA), a dynamic context-aware relay placement algorithm that does not require any prior knowledge of the environment. We developed an open-source simulator for multi-robot expeditions which we used to test both algorithms against state-of-the-art algorithms. Using both Atlas and CARA results in a dynamic context-aware multi-robot expedition that autonomously builds a map of a fully unknown environment, while dynamically placing relays when needed to maintain connectivity that outperforms state-of-the-art algorithms, in terms of time to completion, by a factor of 10.