The primary goal of this handbook is not to present a catalog of isolated techniques, but to chart an emerging research ecosystem centered on spatial intelligence. As established in earlier chapters, spatial intelligence is neither a single method nor a narrowly scoped task. Rather, it is a unifying paradigm that connects representation, perception, reasoning, generation, and action through structured models of the world.
From Isolated Methods to an Integrated Landscape. Much of the existing literature on spatial intelligence is fragmented. Some surveys focus deeply on individual subfields, such as SLAM, neural rendering, or 3D generation, while others adopt a high-level perspective that lacks technical grounding. This handbook aims to bridge this gap by organizing spatial intelligence as an interconnected landscape, where advances in one component are explicitly related to others. Our objective is to reveal how seemingly distinct research threads jointly contribute to world-centered intelligence.
Spatial Intelligence as a Connective Substrate. A central thesis of this handbook is that spatial intelligence functions as a connective substrate for next-generation AI systems. It provides the structural glue that links perception with reasoning, language with geometry, and planning with embodied action. By framing spatial intelligence in this way, the handbook draws connections across computer vision, robotics, simulation, geospatial systems, cognitive science, and neuro-inspired AI. This perspective enables a shared conceptual vocabulary through which contributions from different communities can be meaningfully aligned.
Identifying Gaps and Open Tensions. Beyond synthesizing existing work, this handbook deliberately highlights unresolved gaps in the current research landscape. These include the disconnect between symbolic and neural spatial representations, the limited grounding of perception in actionable world models, and the difficulty of integrating spatial reasoning into long-horizon embodied agents. By making these tensions explicit, we aim to clarify where current paradigms fall short and where future breakthroughs are most likely to emerge.
A Blueprint for an Open Research Program. Ultimately, this handbook is intended as a blueprint for an open and evolving research program. Rather than proposing a closed taxonomy or a definitive solution, it offers a structured lens through which spatial intelligence can be studied, extended, and reimagined. We invite readers not only to learn from the systems and methods surveyed here, but also to rethink spatial intelligence as a first-class organizing principle in the design of generalist intelligent agents.