# FutureMapping: The Computational Structure of Spatial AI Systems

**Authors:** Andrew J. Davison  
**arXiv:** [1803.11288](https://arxiv.org/abs/1803.11288)  
**Date:** 2018-03-29  
**Summary page:** https://www.summarizepaper.com/en/paper/1803.11288v1/

## Key points

- The paper discusses the evolution of Simultaneous Localisation and Mapping (SLAM) in developing a 'Spatial AI' perception capability for intelligent devices
- Emphasizes the significant gap between current visual perception performance and requirements for devices like augmented reality eyewear or consumer robots
- Highlights the need for co-designing algorithms, processors, and sensors to bridge this gap
- Examines the computational structure of current and future Spatial AI algorithms, optimizing them for demanding visual perception tasks
- Stresses the importance of synergy between software and hardware components for cutting-edge Spatial AI capabilities

## Summary

, , , , 

The paper "FutureMapping: The Computational Structure of Spatial AI Systems" by Andrew J. Davison discusses the evolution of Simultaneous Localisation and Mapping (SLAM) in the development of a general geometric and semantic 'Spatial AI' perception capability for intelligent embodied devices. The author highlights the significant gap between current visual perception performance and the requirements of devices like augmented reality eyewear or consumer robots, emphasizing the need for co-designing algorithms, processors, and sensors to bridge this gap. The paper delves into the computational structure of both current and future Spatial AI algorithms, optimizing them to meet demanding visual perception tasks. It also considers algorithmic advancements within ongoing hardware developments, stressing the importance of synergy between software and hardware components for cutting-edge Spatial AI capabilities. Overall, Davison's work provides valuable insights into the potential trajectory of Spatial AI systems, addressing key challenges and opportunities in advancing perceptual capabilities for intelligent devices in various applications such as augmented reality and robotics.