COLMAR: Cooperative View Policy Learning for Multi-Agent Active 3D Reconstruction

2026-07-15

AI Summary

This work addresses the inefficiency in collaborative multi-agent active 3D reconstruction caused by redundant observations and spatial clustering. To overcome this, the authors propose a communication-free cooperative viewpoint selection strategy that guides team-level exploration, overlapping coverage, and collision-free navigation through a reconstruction-aware objective. Agent behaviors are aligned via a shared policy network and incremental reconstruction feedback, ensuring consistent exploration and high geometric fidelity. The policy is trained using a parameter-sharing proximal policy optimization (PPO) algorithm and, at deployment, each agent independently selects actions based solely on a fused occupancy map. High-fidelity reconstruction is achieved by integrating 3D Gaussian Splatting. Evaluated on the GLEAM and Replica datasets, the method improves reconstruction accuracy by up to 54% and coverage by up to 49% over heuristic and non-collaborative baselines under identical sensing budgets.

Abstract

Active 3D reconstruction requires selecting informative viewpoints under limited sensing budgets. In multi-agent settings, coordination inefficiencies such as redundant observations and spatial clustering can significantly reduce reconstruction quality. We present COLMAR, a cooperative view policy learning framework for multi-agent active 3D reconstruction. COLMAR formulates viewpoint allocation as a shared policy optimization over map-centric observations and introduces a reconstruction-aware objective that promotes overlap-aware coverage, team-level discovery, and collision-safe exploration. Dense feedback derived from incremental reconstruction updates aligns exploration behavior with downstream geometric quality. The policy is trained using parameter-sharing Proximal Policy Optimization (PPO) with independent per-agent action selection at deployment, conditioned on a fused team map and without inter-agent message passing for decision making. Selected viewpoints are then reconstructed with 3D Gaussian Splatting (3DGS) for high-fidelity photometric evaluation. Experiments on GLEAM and Replica demonstrate consistent improvements over heuristic and non-cooperative baselines, achieving up to 54% higher reconstruction accuracy and 49% greater coverage under matched sensing budgets.

Problem

• multi-agent
• active 3D reconstruction
• viewpoint selection
• coordination inefficiency
• sensing budget

Innovation

• cooperative multi-agent
• active 3D reconstruction
• view policy learning
• reconstruction-aware objective
• 3D Gaussian Splatting