Core Concepts
Investigating task-driven exploration with mobile sensors using intelligent map compression to optimize robot path-planning.
Abstract
This paper explores task-driven exploration of unknown environments with mobile sensors communicating compressed measurements. It introduces a novel communication framework and a multi-agent exploration algorithm. The study focuses on optimizing actions based on compression uncertainty, reducing time to reach the target without overloading the communication network.
Directory:
Introduction
Rise in autonomous robot teams.
Importance of effective communication.
Preliminaries: Grid World and Abstractions
Representation of environment by 2D occupancy grid.
Definition of abstraction for compressed representation.
Communication of Abstracted Environments
Robots communicate abstractions to optimize performance.
Problem Formulation
Team of robots navigating through unfamiliar environment.
Framework Architecture
Components: Decoder, Path Planner, Sensor Action Selector, Path Converter, Encoder.
Experiments
Two scenarios tested in a 64x64 environment.
Conclusions
Study on exploring unknown environments with mobile sensors.
Stats
"The simulations involved a single Sensor."
"We set nm = 24, and ni = 4."
"The Seeker’s sensing region is a 5 × 5 grid around its position."