On the Application of Efficient Neural Mapping to Real-Time Indoor Localisation for Unmanned Ground Vehicles

TL;DR

This paper demonstrates that a constrained 2D neural mapping approach can enable real-time, centimeter-accurate indoor localisation for unmanned ground vehicles using visual data, suitable for embedded platforms.

Contribution

It introduces a compact neural localisation model optimized for real-time inference on embedded systems, with a large dataset and deployment on a UGV platform.

Findings

Achieves 9cm mean localisation accuracy in real-world tests

Runs at 6fps on onboard CPU, 35fps on embedded GPU, 220fps on desktop GPU

Provides a new dataset with simulated and real environment samples

Abstract

Global localisation from visual data is a challenging problem applicable to many robotics domains. Prior works have shown that neural networks can be trained to map images of an environment to absolute camera pose within that environment, learning an implicit neural mapping in the process. In this work we evaluate the applicability of such an approach to real-world robotics scenarios, demonstrating that by constraining the problem to 2-dimensions and significantly increasing the quantity of training data, a compact model capable of real-time inference on embedded platforms can be used to achieve localisation accuracy of several centimetres. We deploy our trained model onboard a UGV platform, demonstrating its effectiveness in a waypoint navigation task, wherein it is able to localise with a mean accuracy of 9cm at a rate of 6fps running on the UGV onboard CPU, 35fps on an embedded GPU, or 220fps on a desktop GPU. Along with this work we will release a novel localisation dataset comprising simulated and real environments, each with training samples numbering in the tens of thousands.