HGS-Planner: Hierarchical Planning Framework for Active Scene Reconstruction Using 3D Gaussian Splatting

TL;DR

This paper introduces HGS-Planner, a hierarchical planning framework utilizing 3D Gaussian Splatting for fast, high-quality active scene reconstruction in robotics, improving real-time performance and scene understanding.

Contribution

It presents a novel hierarchical planning approach that adaptively guides scene reconstruction using 3D Gaussian Splatting, combining global and local planning for efficiency.

Findings

Outperforms existing real-time reconstruction methods

Demonstrates effectiveness in simulated environments

Validates approach in real-world scenarios

Abstract

In complex missions such as search and rescue,robots must make intelligent decisions in unknown environments, relying on their ability to perceive and understand their surroundings. High-quality and real-time reconstruction enhances situational awareness and is crucial for intelligent robotics. Traditional methods often struggle with poor scene representation or are too slow for real-time use. Inspired by the efficacy of 3D Gaussian Splatting (3DGS), we propose a hierarchical planning framework for fast and high-fidelity active reconstruction. Our method evaluates completion and quality gain to adaptively guide reconstruction, integrating global and local planning for efficiency. Experiments in simulated and real-world environments show our approach outperforms existing real-time methods.