IntentionNet: Map-Lite Visual Navigation at the Kilometre Scale

TL;DR

IntentionNet is a neural network-based navigation system enabling robots to traverse indoor and outdoor environments over kilometre distances using intentions, demonstrating robustness to errors and successful deployment on a Boston Dynamics Spot robot.

Contribution

The paper introduces IntentionNet, a novel neural network architecture for scalable, robust navigation using intentions, and demonstrates its effectiveness in real-world kilometre-scale outdoor and indoor navigation.

Findings

Successfully navigates over a kilometre in complex environments

Robust to significant metric positioning and mapping errors

Deployed on a Boston Dynamics Spot robot

Abstract

This work explores the challenges of creating a scalable and robust robot navigation system that can traverse both indoor and outdoor environments to reach distant goals. We propose a navigation system architecture called IntentionNet that employs a monolithic neural network as the low-level planner/controller, and uses a general interface that we call intentions to steer the controller. The paper proposes two types of intentions, Local Path and Environment (LPE) and Discretised Local Move (DLM), and shows that DLM is robust to significant metric positioning and mapping errors. The paper also presents Kilo-IntentionNet, an instance of the IntentionNet system using the DLM intention that is deployed on a Boston Dynamics Spot robot, and which successfully navigates through complex indoor and outdoor environments over distances of up to a kilometre with only noisy odometry.