Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Distributed Optimization with Consensus Constraint for Multi-Robot Semantic Octree Mapping

Arash Asgharivaskasi, Nikolay Atanasov

TL;DR

This work addresses distributed semantic mapping for multi-robot teams under one-hop communication constraints. It represents per-cell class distributions with a log-odds vector $h$ in R^(C+1) and optimizes the sum of local log-likelihoods with a consensus penalty that aligns neighboring robots’ maps, enabling decentralized fusion. A semantic octree is employed to compress the multi-class map, achieving memory and bandwidth efficiency while preserving accuracy. Experiments in a Unity environment with six robots demonstrate convergence to a globally consistent map and substantial communication savings, supporting decentralized planning without a central estimator.

Abstract

This work develops a distributed optimization algorithm for multi-robot 3-D semantic mapping using streaming range and visual observations and single-hop communication. Our approach relies on gradient-based optimization of the observation log-likelihood of each robot subject to a map consensus constraint to build a common multi-class map of the environment. This formulation leads to closed-form updates which resemble Bayes rule with one-hop prior averaging. To reduce the amount of information exchanged among the robots, we utilize an octree data structure that compresses the multi-class map distribution using adaptive-resolution.

Distributed Optimization with Consensus Constraint for Multi-Robot Semantic Octree Mapping

TL;DR

This work addresses distributed semantic mapping for multi-robot teams under one-hop communication constraints. It represents per-cell class distributions with a log-odds vector in R^(C+1) and optimizes the sum of local log-likelihoods with a consensus penalty that aligns neighboring robots’ maps, enabling decentralized fusion. A semantic octree is employed to compress the multi-class map, achieving memory and bandwidth efficiency while preserving accuracy. Experiments in a Unity environment with six robots demonstrate convergence to a globally consistent map and substantial communication savings, supporting decentralized planning without a central estimator.

Abstract

This work develops a distributed optimization algorithm for multi-robot 3-D semantic mapping using streaming range and visual observations and single-hop communication. Our approach relies on gradient-based optimization of the observation log-likelihood of each robot subject to a map consensus constraint to build a common multi-class map of the environment. This formulation leads to closed-form updates which resemble Bayes rule with one-hop prior averaging. To reduce the amount of information exchanged among the robots, we utilize an octree data structure that compresses the multi-class map distribution using adaptive-resolution.
Paper Structure (4 sections, 5 equations, 4 figures, 1 algorithm)

This paper contains 4 sections, 5 equations, 4 figures, 1 algorithm.

Table of Contents

  1. Introduction
  2. Problem Statement
  3. Distributed Semantic Mapping
  4. Evaluation

Figures (4)

  • Figure 1: Semantically annotated point cloud obtained from a range and category observation $z_t^i$. Each object category is shown with a unique color.
  • Figure 2: Time lapse of multi-robot autonomous exploration and semantic octree mapping in the environment shown in (d). Different colors represent different semantic categories (building, dirt road, grass, etc.). Local semantic maps are overlaid with each other such that the transparency of each cell directly relates to the deviation between the local map estimates.
  • Figure 3: Time evolution of the total distance in local map estimates. Iteration of Alg. \ref{['alg:dist_mapping']} is executed in each robot every one second, upon arrival of a local map estimate from a neighboring robot.
  • Figure 4: Packet size used for local map communication for uniform resolution grid map (blue) and semantic octree map (red). The solid lines and the dashed lines show the average over all robots and one standard deviation from the average, respectively.