Imagine a team of robots navigating a cluttered house, not bumping into each other, and efficiently finding everything you need. That's the promise of CAMON, a new framework that lets robots chat their way to collaborative success. Traditionally, robots have struggled to work together on complex navigation tasks. Single robots get lost or take too long, and coordinating multiple robots is a programming nightmare. CAMON tackles this challenge by enabling robots to have LLM-powered conversations, similar to how humans discuss plans. Each robot builds a semantic understanding of its surroundings, including identifying rooms and objects within them. Instead of blindly wandering, a robot entering a new room can ask the 'team leader' which objects it should focus on, preventing duplicated effort and conflicts. This dynamic leadership structure is key—the role of leader shifts between robots based on who has the most up-to-date global information. This makes the system resilient: if one robot malfunctions, the team can carry on without missing a beat. CAMON isn’t just about communication. It combines this with smart mapping and path planning. Robots build topological maps to understand the layout and use efficient algorithms to plot their routes. When a robot finds a new room, it captures images and uses GPT-4o to generate descriptions, enriching the team’s shared knowledge. This combination of communication, perception, and planning makes CAMON highly effective in multi-object navigation. While the current version of CAMON is limited to single-floor navigation and struggles with dynamic obstacles like pets, it paves the way for more sophisticated robot teams. Future research aims to integrate manipulation tasks, bringing us closer to a world where robots seamlessly collaborate to accomplish complex goals, whether it's tidying up the house or assisting in search and rescue missions.