Multi-agent Deployment Problem

07 Aug 2020

Left: Movement of 10 robots through unknown, nonconvex space. Right: Evolution of the density function over time - darker being more dense. You can see that the robots eventually establish patterns of movement.

For my ES 159: Introduction to Robotics class, I wanted to look at a SLAM algorithm that would explore a space while also maintaining observational capabilities in that space - what is termed the deployment problem. I found a paper by Kantaros et al., which addressed a known version of the problem.

With help from Yiannis Kantaros at UPenn, I was able to replicate their results, as well as demonstrating feasibility in unknown environments. In addition, I added patrol behavior by using time-varying density functions to modify robot motion, which has interesting theoretical and practical considerations. You can see the draft here

I’m currently working on expanding this concept into a submitted paper for review - more news on this soon!

Skills: MATLAB, Controls, Robotics

Victor Qin

Multi-agent Deployment Problem

Left: Movement of 10 robots through unknown, nonconvex space. Right: Evolution of the density function over time - darker being more dense. You can see that the robots eventually establish patterns of movement.