Augmented Reality Support for Swarm Monitoring: Evaluating Visual Cues to Prevent Fragmentation

University of Brest
3 to 6 months
35h / week
English B2

Historically based on models inspired by natural phenomena (flocks of birds, particle physics, social insects, etc.), robot swarms are groups of autonomous robots that self-organise locally and are capable of performing certain tasks more efficiently than a single robot. Among multi-robot systems, swarms are distinguished by the way their individuals collaborate autonomously, through local interactions and without knowledge of the overall situation or task of the swarm. Robot swarms are so-called complex systems, composed of a large number of autonomous entities sharing the same environment and interacting with each other locally without any centralised control. This mode of operation promotes the emergence of macroscopic properties such as robustness, adaptability and scalability. Robot swarms can exhibit self-organised collective behaviours, such as flocking. The ARTUISIS project aims to facilitate interaction between humans and swarms by designing augmented visualisations in Virtual Reality to improve operators’ understanding. The very source of these properties, the emergence of the swarm’s behaviour through the interactions between the robots, has been preventing operators from keeping a mental model of the swarm behaviour over time. The operator role is involved in monitoring the swarm and compensating for limitations in the swarm’s autonomy, such as fragmentation ( when the swarm loses its cohesion and robots separate from it). When human perceptions of the swarm do not allow them to understand what is happening, one solution is to provide them with the missing information they need, for example through data visualisations. Information can take the form of different visualisations, and some forms may be more suitable than others depending on the situation. In the ARTUISIS project, we chose to study visualisations adapted for communication using an Augmented Reality device, and more specifically visualisations spatially located on the swarm and its robots. Initial work proposed by Aymeric Hénard, a PhD student working on the project, made it possible to evaluate, in a virtual reality environment, the ability of four localised visualisations to improve human anticipation of swarm fragmentation and to choose the appropriate control to prevent it from occurring. As the results of this initial assessment did not identify any visualisation as being more suitable than the others, the overall objective of this internship is to make new proposals, implement them and evaluate them.

Internship allowance : 600 euros per month as per national law

Tasks and duties entrusted to the student:
  • Providing an overview of complex systems and their visualisation in the litterature  Identify the information to be transmitted to an operator based on robot behaviours and human perception
  • Propose and implement localised and augmented visualisation metaphors of this information
  • Define a protocol for evaluating these visualisations
  • Evaluate these visualisations with subjects
Skills to be acquired or developed:
  • Carry out a state of the art associated with a scientific question
  • Write a scientific article according to international standards
  • AR and VR techniques and technologies
  • Data visualisation, 3D metaphors
  • Distributed and self-organisation algorithms
  • Conduct a user study
Jérémy Rivière / jeremy.riviere@univ-brest.fr / Lab-STICC, UMR CNRS 6285 Immersive Natural User Interaction Team (INUIT)