Path Planning for Resilient Multi-robot Connectivity

Robotics and automation are radically transforming industries—from manufacturing to services—as well as societies.
Collaborative multi-robots, in particular, have the potential to greatly improve the performance of any spatial task (e.g., exploration and surveillance) much like distributed computing impacted information science.
One fundamental, enabling property of these systems is the ability for their members to exchange information with one another.
Until today, researchers have often combined sophisticated “connectivity maintenance” approaches (i.e., algebraic connectivity, spring-damper models, etc) with relatively trivial exploration strategies.
In this project, we intend to merge our home institution’s expertise in multi-robot connectivity with the host institution’s research on multi-robot path planning.
We will also pay special attention to the resilience of our methodology to cope with the non-idealities of a real-world deployment.
In our vision, this project will create a multi-robot system capable of autonomously implement a complex exploration strategy that respects connectivity constraints, even in the presence of hardware and software errors.
Our deliverables will include: (i) theory for a new robotic controller, (ii) its implementation as multi-(robotic-)platform software, and (iii) the analysis of an experimental campaign in actual robotic hardware.

Faculty Supervisor:

Giovanni Beltrame

Student:

Partner:

University of Cambridge

Discipline:

Computer science

Sector:

Education

University:

École Polytechnique de Montréal

Program: