Autonomous Robot for Disaster Mapping and Victim Localization

TL;DR

The work tackles autonomous disaster reconnaissance by integrating mapping with victim localization on a TurtleBot3 using ROS Noetic. It combines an enhanced exploration strategy based on frontier detection and next-best-view concepts with a Cubature Kalman Filter-based 3D tag pose estimator (CKF3D) and a dedicated search-and-rescue routine to ensure comprehensive map coverage and accurate AprilTag localization. Results in Gazebo and real-world arenas show robust mapping and tag detection, with CKF3D improving localization accuracy in some settings and the custom exploration yielding faster map coverage than a baseline frontier method, while highlighting challenges such as sensing occlusions, tag-size biases, and hardware reliability. The approach advances practical autonomous search-and-rescue by delivering an integrated pipeline, experimental validation, and actionable insights for deployment in dynamic disaster environments.

Abstract

In response to the critical need for effective reconnaissance in disaster scenarios, this research article presents the design and implementation of a complete autonomous robot system using the Turtlebot3 with Robotic Operating System (ROS) Noetic. Upon deployment in closed, initially unknown environments, the system aims to generate a comprehensive map and identify any present 'victims' using AprilTags as stand-ins. We discuss our solution for search and rescue missions, while additionally exploring more advanced algorithms to improve search and rescue functionalities. We introduce a Cubature Kalman Filter to help reduce the mean squared error [m] for AprilTag localization and an information-theoretic exploration algorithm to expedite exploration in unknown environments. Just like turtles, our system takes it slow and steady, but when it's time to save the day, it moves at ninja-like speed! Despite Donatello's shell, he's no slowpoke - he zips through obstacles with the agility of a teenage mutant ninja turtle. So, hang on tight to your shells and get ready for a whirlwind of reconnaissance! Full pipeline code https://github.com/rzhao5659/MRProject/tree/main Exploration code https://github.com/rzhao5659/MRProject/tree/main

Figures (12)

• Figure 1: Gazebo simulation arenas. (a) is 15.0m$\times$10.0m (b) is 3.0m$\times$3.0m (c) is 34.5m$\times$21.75m.
• Figure 2: Example real-world arena setup. Please note that the final setup is larger and includes more AprilTags, as depicted in the demo videos.
• Figure 3: Nodes and topics in our solution
• Figure 4: Expanding Wavefront Frontier Detection. Blue cells are frontier cells from previous timestep. Red cells are new, unexplored free cells that are traversed and evaluated as potential frontier cells. Figure from approaches-frontiers.
• Figure 5: Computation of exploration goal. a) Sampling process b) Potential gain. The yellow arrows indicate best pose near each frontier, and the green arrow indicate the best pose overall.