Robotics Design, Repair and Restoration
Swarm Robotics Management within Disaster Response
Revisiting my Bachelors year thesis project
In this article, we revisit my univeristy bachelours project "Integrating Object Recognition in Swarm Robotics Management" of 2022.
In a world with so many prominent large-scale disasters, I envisioned a world where such scenarios could be resolved without adding risk to any subsequent human life.
This concept is what I sought to explore during my Computer Systems Engineering bachelor's thesis back in 2022.
Introduction
1 - The Concept
My project, "Integrating Object Recognition in Swarm Robotics Management", aimed to create a proof-of-concept system designed to perform cleanup operations within a disaster scenario using cheep, simplistic, and autonomous robots.
Within disaster scenarios, vast varieties of time-critical hazards must be resolved. Unfortunately, sending in human responders to save already exposed civilians puts the responders themselves in danger, with possibly fatal ramifications. Considering the existing use of robotics in hazardous situations (e.g. bomb disposal, mineshaft exploration, extraterrestrial planet surveys and space exploration), could we further expand autonomous robotics within this domain, limiting the need for on-site personnel? Or even limit human intervention altogether?
My initial investigation showed that existing radio-controlled systems, such as those previously used in Chernobyl, required on-site operators. However, just as back then, putting a competent computer system intelligent enough to perform tasks to the level required did not seem possible. This suggested that another approach, cheaper, whilst also resilient to danger, needed to be explored.
But how can we give the robots the benefits that come with AI without actually giving the robots themselves AI?
A tricky question at the start, but what if the AI did not have to reside entirely within each individual robot intended to be used? That would certainly reduce each robot's computing overhead and, therefore, the per unit hardware implementation costs.