Spatial Coordinate Transformation for 3D Neural Implicit Mapping
Kyeongsu Kang, Seongbo Ha, Sibaek Lee, Hyeonwoo Yu
AI summary
Problem
INR-based SLAM encodes maps in neural network weights, making post-loop-closure remapping computationally expensive and memory-intensive due to required retraining or complex submap architectures.
Approach
Approximates a nonlinear spatial coordinate transformation between pre- and post-optimization domains using Gaussian Process regression, resolves mapping ambiguities with a temporal domain, and removes it via uncertainty estimation to update the map without retraining.
Key results
- Eliminates post-training requirement for map updates
- Significantly reduces memory and space complexity
- Resolves one-to-many coordinate mapping ambiguities
- Validates effective remapping on indoor datasets
Why it matters
Enables scalable, real-time INR-based SLAM on resource-constrained robotics platforms by drastically cutting loop-closure computational overhead.
Abstract
Implicit Neural Representation (INR)-based SLAM has a critical issue where all keyframes must be stored in memory for post-training whenever a remapping is needed due to the neural network’s weights themselves representing the map. To address this, previous INR-based SLAM proposed methods to modify INR-based maps without changing the neural network’s weights. However, these approaches suffer from low memory efficiency and increased space complexity. In this letter, we introduce a remapping method for INR-based maps that does not require post-training the neural network’s weights and needed low space cost. The problem of function modification, such as updating a map defined as a neural network function, can be viewed as transforming the function’s domain. Leveraging function domain transformation, we propose a method to update INR-based maps by identifying the transforma- tion function between the post-optimization and pre-optimization domains. Additionally, to prevent cases where the transformation between the post-optimization and pre-optimization domains does not form a one-to-many relationship, we introduce a temporal do- main and propose a method to find the spatial coordinate transfor- mation function accordingly. Evaluations in INR-based techniques demonstrate that our proposed method effectively update to maps while requiring significantly less memory compared to existing remapping approaches.