Pre-Publication V2. AI systems that perceive, are designed to analyze, represent and predict, based on available data, interpret, and predict the geometric and physical relationships between objects, spaces, and agents in two- or three-dimensional spaces environments. These models understand properties such as depth, distance, orientation, scale, occlusion, and motion dynamics, enabling applications in robotics, autonomous navigation, virtual production, simulation, and augmented or mixed reality. Spatial reasoning may operate on real-world sensor input, synthetic environments, or both.
Deliberation Summary:
Consider utilizing language that better describes what the system does, is more objective and not overly broad, particularly in the context of the definition.
Pre-Publication V1. AI systems that perceive, interpret, and predict the geometric and physical relationships between objects, spaces, and agents in two- or three-dimensional environments. These models understand properties such as depth, distance, orientation, scale, occlusion, and motion dynamics, enabling applications in robotics, autonomous navigation, virtual production, simulation, and augmented or mixed reality. Spatial reasoning may operate on real-world sensor input, synthetic environments, or both.