Core Concepts
Collective exploration and rendezvous algorithms for two mobile agents in unknown graphs improve over naive strategies.
Abstract
The content discusses exploration and rendezvous problems in mobile computing, focusing on two agents moving along edges. It introduces algorithms improving competitive ratios for graph exploration and meeting as fast as possible. The article covers historical context, depth-first search, asynchronous navigation models, collective tree exploration, weighted graphs, continuous moves, and generalizations.
Historical Context:
Shannon's electromechanical mouse inspired micro-mouse competitions globally.
Depth-first search algorithm dates back to the late 19th century.
Depth-First Search:
Trémaux's algorithm ensures traversal of all edges once in both directions.
Collective Exploration:
Algorithms aim to traverse all edges collectively with improved competitive ratios.
Rendezvous Problem:
Algorithms guarantee meeting within a specified number of time-steps with improved efficiency.
Navigation Model:
Agents move through steps involving reading, marking, traversing edges, and backtracking if needed.
Weighted Graphs:
Algorithms extend to weighted graphs while maintaining cost-efficient traversal guarantees.
Continuous Moves:
The model adapts to continuous agent movements controlled by an adversary while ensuring efficient exploration or rendezvous.
Stats
Depth-first search generalizes the ancient maze-solving heuristic 'right-hand-on-the-wall'.
The algorithm achieves graph exploration in 2m moves where m is the number of edges of the graph.
The best competitive ratio for collective exploration is O(k/log(k)) with distributed communication or O(√k) with complete communication.
The 'Wait for Mommy' algorithm requires 2m time-steps for rendezvous between two agents in an unknown graph.
Quotes
"In this paper, we start by studying the question of whether two agents initially located at the same node can solve a maze faster than a single agent."
"Our contribution to this problem is an algorithm achieving rendezvous of two mobile agents in only ⌈3/2m⌉ time-steps."
"Our guarantees hold for a general model of asynchrony and generalize well to weighted graphs."