Niantic is developing a Large Geospatial Model (LGM) that utilizes player-contributed scans of real-world locations to enhance augmented reality (AR) experiences. By employing machine learning and over 50 million neural networks, the LGM aims to help computers understand and navigate physical spaces similarly to how humans perceive them. This model captures detailed 3D information about locations, allowing for intelligent extrapolation of unseen areas based on previously scanned data. The LGM’s capabilities will facilitate new AR applications, enhance user interactions, and support various fields such as logistics and design. As wearable technology like AR glasses becomes more common, the integration of LGMs into everyday life could revolutionize how we experience and interact with our surroundings.
| name | description | change | 10-year | driving-force | relevancy |
|---|---|---|---|---|---|
| User-Contributed Scans | Players scan public locations to enhance AR experiences and train AI models. | Shifting from passive interaction to active contribution in AR development. | AR experiences will be heavily personalized and community-driven, enhancing user engagement. | Growing user participation in content creation and the desire for immersive experiences. | 4 |
| Spatial Intelligence in AI | AI is developing spatial understanding akin to human perception of environments. | Transitioning from basic visual recognition to advanced spatial reasoning in AI. | AI will accurately interpret and interact with physical spaces, enabling seamless AR integration. | Advancements in machine learning techniques and increased data availability. | 5 |
| Large Geospatial Models (LGM) | Models that understand physical spaces through vast amounts of visual data. | From localized models to global understanding and contextual awareness in AI. | AI will possess a comprehensive understanding of global geographies and their interrelations. | The necessity for enhanced navigation and interaction in complex real-world environments. | 5 |
| Integration of Foundation Models | Emerging models will communicate with each other for comprehensive understanding. | Moving from isolated AI models to interconnected systems for better insights. | A cohesive ecosystem of AI models will revolutionize spatial computing applications. | The complexity of real-world applications requiring collaborative model capabilities. | 4 |
| AR Glasses and Wearable Technology | AR glasses are anticipated to become mainstream in the future. | Transitioning from smartphones to wearable technology for real-world interaction. | Wearable AR technology will enhance real-world navigation and user engagement significantly. | Consumer demand for more interactive and immersive tech experiences. | 4 |
| Persistent Digital Content | Users can place and share digital content in physical locations. | Evolving from temporary digital experiences to permanent and shareable content. | Digital interactions will become a staple of physical spaces, influencing urban design. | The desire for social interaction and engagement through shared digital experiences. | 3 |
| User-Centric AR Development | AR experiences are increasingly shaped by user input and behavior. | From developer-driven content to user-driven experiences in augmented reality. | AR applications will become highly personalized, adapting to individual user needs and preferences. | The shift towards user empowerment and community involvement in technology. | 4 |
| Cultural and Geographical Context in AI | AI learns cultural norms and geographical characteristics for better understanding. | From generic AI responses to context-aware interactions based on location and culture. | AI will provide tailored experiences that respect and reflect local cultures and practices. | The need for culturally sensitive technology in a globalized world. | 4 |
| name | description | relevancy |
|---|---|---|
| Privacy and Surveillance | Widespread scanning and data collection from public locations could lead to privacy concerns and surveillance issues for individuals in these areas. | 4 |
| Data Security and Misuse | The volumetric data collected from user scans may be susceptible to breaches or misuse, raising concerns about data protection. | 5 |
| Dependence on Technology | As reliance on AR and geospatial models increases, there is a risk of becoming overly dependent on technology for navigation and spatial understanding. | 4 |
| Inequitable Access to Technology | The development of AR technologies could create a digital divide, leaving behind those who lack access to the necessary devices or internet connectivity. | 3 |
| Cultural Misinterpretations | Geospatial models driven by AI may misinterpret or inadequately represent cultural landmarks, leading to potential miscommunication or disrespect. | 3 |
| Job Displacement | As AR technologies and geospatial systems advance, there might be displacement of jobs in traditional navigation and tourism roles. | 4 |
| name | description | relevancy |
|---|---|---|
| User-Contributed Scanning | Encouraging users to scan real-world locations to enhance AR experiences and improve geospatial data models. | 5 |
| Integration of 3D and Geospatial Intelligence | Combining 3D data models with geospatial intelligence for better understanding of physical spaces. | 5 |
| Spatial Computing Transition | Shifting from mobile to wearable technology for more immersive AR experiences. | 5 |
| Persistent AR Content | Creating AR content that remains in specific locations for user interaction, enabling shared experiences. | 4 |
| Foundation Models Interconnectivity | Developing complementary AI models (LLMs, multimodal models, LGMs) for enhanced understanding of the world. | 4 |
| Human-Like Spatial Understanding in AI | Advancing AI’s ability to perceive and understand physical spaces in a human-like manner. | 5 |
| Real-Time Navigation and Guidance | Utilizing geospatial models for real-time guidance and personalized recommendations in physical environments. | 4 |
| Cross-Application of Geospatial Models | Applying Large Geospatial Models beyond gaming to fields like logistics, urban planning, and remote collaboration. | 5 |
| description | relevancy | src |
|---|---|---|
| A model using large-scale machine learning to understand and connect scenes globally, enabling advanced spatial intelligence. | 5 | 65e4c914b47f88920bb9442d4d102b67 |
| A system allowing devices to determine position and orientation using 3D maps built from user-contributed scans. | 5 | 65e4c914b47f88920bb9442d4d102b67 |
| The ability for machines to perceive and understand physical spaces, crucial for AR applications and robotics. | 5 | 65e4c914b47f88920bb9442d4d102b67 |
| Models that interpret and create static and moving images, evolving from 2D to 3D understanding. | 4 | 65e4c914b47f88920bb9442d4d102b67 |
| Maps represented by neural networks that encode location data and provide positioning with centimeter-level accuracy. | 4 | 65e4c914b47f88920bb9442d4d102b67 |
| Versatile AI models that interact across different modalities, enhancing understanding of the physical world. | 4 | 65e4c914b47f88920bb9442d4d102b67 |
| Wearable technology that integrates augmented reality into daily life, relying on spatial computing advancements. | 5 | 65e4c914b47f88920bb9442d4d102b67 |
| The capability of understanding and manipulating 3D representations of the world for enhanced AR experiences. | 4 | 65e4c914b47f88920bb9442d4d102b67 |
| A neural network that positions camera views relative to each other, even under drastic viewpoint changes. | 3 | 65e4c914b47f88920bb9442d4d102b67 |
| name | description | relevancy |
|---|---|---|
| Spatial Intelligence in AI Models | The development of AI models that can understand and interact with physical spaces, marking a shift from text and 2D to 3D data interpretation. | 5 |
| Integration of Geospatial Models with Wearable Technology | The potential for Large Geospatial Models to enhance AR glasses and other wearable tech, providing real-time interaction with physical environments. | 5 |
| User-Contributed Geospatial Data | The reliance on user-contributed scans for building comprehensive geospatial models, raising questions about data privacy and ownership. | 4 |
| Cultural and Geographical Adaptation of AI | The need for AI models to understand cultural contexts and geographic variations in physical structures and interactions. | 4 |
| Foundation Models Interoperability | The emerging need for different types of AI models (LLMs, multimodal models, LGMs) to work together for comprehensive understanding of the world. | 5 |
| Applications Beyond Gaming | Expanding applications of Large Geospatial Models in fields like logistics, urban planning, and remote collaboration, beyond gaming environments. | 5 |
| Data Scarcity and Model Limitations | The challenges posed by limited local data and the need for extensive geospatial data to improve model accuracy and applicability. | 4 |