Imagine a robot seamlessly navigating a cluttered home, effortlessly finding your misplaced keys or fetching a specific book from a shelf it's never seen before. This is the exciting promise of open-world object-goal navigation (OWON). Researchers are tackling this complex challenge by building systems that mimic human reasoning and spatial understanding, allowing robots to search for and locate objects in completely new environments. The key innovation lies in the use of Open Scene Graphs (OSGs), dynamic maps that organize scene information in a way that makes sense to AI. Unlike traditional maps that are fixed and environment-specific, OSGs adapt to different spaces, capturing the relationships between objects and regions within a scene. This adaptability allows robots to generalize their knowledge and navigate novel spaces zero-shot, without prior training on the specific environment. This is achieved by combining powerful AI building blocks. Visual Foundation Models (VFMs) identify objects and regions within a scene from visual input. Large Language Models (LLMs) use this information to reason about object locations, drawing upon their vast knowledge about how human environments are typically organized. General Navigation Models (GNMs) then translate the LLM’s plans into actual movement commands for the robot. OpenSearch, a prototype system built with this approach, has shown promising results in both simulation and real-world tests. It successfully navigates diverse environments, including homes and supermarkets, locating specified objects even with open-vocabulary, natural language commands. However, current systems face real-world challenges. Processing information and making decisions can be slow due to the complexity of the AI models. Improving efficiency and enabling real-time navigation will be crucial for practical applications. Additionally, handling uncertainty in perception and decision-making is an area of ongoing research. As the field of robotics continues to evolve, breakthroughs in OWON research will pave the way for more versatile, adaptable robots capable of performing complex tasks in our homes, workplaces, and beyond.