NeuroMesh: A Unified Neural Inference Framework for Decentralized Multi-Robot Collaboration

2026-04-16

AI Summary

This work addresses the challenges of efficiently deploying learned models on heterogeneous multi-robot systems, which are hindered by hardware disparities, limited communication, and the absence of a unified execution stack. To overcome these issues, the authors propose a modular, decentralized, cross-platform neural inference framework that standardizes observation encoding, message passing, information aggregation, and task decoding. The framework innovatively integrates a dual-aggregation paradigm combining reduction and broadcast communication and features a parallel architecture that decouples cycle time from end-to-end latency, enabling effective heterogeneous collaboration. Implemented in high-performance C++ and integrated with the Zenoh communication middleware, it supports hybrid GPU/CPU inference. Experiments on aerial-ground heterogeneous robot teams demonstrate its effectiveness in collaborative perception, decentralized control, and task allocation, exhibiting strong robustness across diverse task structures and workload scales.

Technology Category

• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Hardware-aware ML
• Computer Vision: Multi-modal Vision

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems

Abstract

Deploying learned multi-robot models on heterogeneous robots remains challenging due to hardware heterogeneity, communication constraints, and the lack of a unified execution stack. This paper presents NeuroMesh, a multi-domain, cross-platform, and modular decentralized neural inference framework that standardizes observation encoding, message passing, aggregation, and task decoding in a unified pipeline. NeuroMesh combines a dual-aggregation paradigm for reduction- and broadcast-based information fusion with a parallelized architecture that decouples cycle time from end-to-end latency. Our high-performance C++ implementation leverages Zenoh for inter-robot communication and supports hybrid GPU/CPU inference. We validate NeuroMesh on a heterogeneous team of aerial and ground robots across collaborative perception, decentralized control, and task assignment, demonstrating robust operation across diverse task structures and payload sizes. We plan to release NeuroMesh as an open-source framework to the community.

Problem

• heterogeneous robots
• multi-robot collaboration
• hardware heterogeneity
• communication constraints
• unified execution stack

Innovation

• NeuroMesh
• decentralized multi-robot collaboration
• neural inference framework
• dual-aggregation paradigm
• heterogeneous robot systems