Stein Variational Belief Propagation for
Multi-Robot Coordination

Jana Pavlasek

University of Michigan

Joshua Mah

University of Michigan

Ruihan Xu

University of Michigan

Odest Chadwicke Jenkins

University of Michigan

Fabio Ramos

NVIDIA Corporation
University of Sydney

Decentralized coordination for multi-robot systems involves planning in challenging, high-dimensional spaces. The planning problem is particularly challenging in the presence of obstacles and different sources of uncertainty such as inaccurate dynamic models and sensor noise. In this paper, we introduce Stein Variational Belief Propagation (SVBP), a novel algorithm for performing inference over nonparametric marginal distributions of nodes in a graph. We apply SVBP to multi-robot coordination by modelling a robot swarm as a graphical model and performing inference for each robot. We demonstrate our algorithm on a simulated multi-robot perception task, and on a multi-robot planning task within a Model-Predictive Control (MPC) framework, on both simulated and real-world mobile robots. Our experiments show that SVBP represents multi-modal distributions better than sampling-based or Gaussian baselines, resulting in improved performance on perception and planning tasks. Furthermore, we show that SVBP's ability to represent diverse trajectories for decentralized multi-robot planning makes it less prone to deadlock scenarios than leading baselines.

Multi-Robot Perception

We implement our SVBP on a graph problem simulating perception under challenging noise conditions. The goal is to localize each node in the "spider" where the observation is the multi-modal Gaussian shown in the video.

SVBP (Ours)

Particle Belief Propagation

SVBP maintains modes of the distribution more effectively than Particle Belief Propagation, which relies on importance sampling.

Multi-Robot Planning

We apply SVBP to the multi-robot planning problem. Below are video results for each scene in the dataset.

Baselines

We implement different baselines to compare our algorithm. Click through the examples for each below.

Gaussian Belief Propagation

A version of our algorithm with a Gaussian trajectory distribution instead of a nonparametric one. The GaBP baseline is more likely to fall into local minima.

ORCA (20 cm radius)

The popular ORCA method (van den Berg 2011) is efficient but not as smooth as SVBP and prone to deadlock.

ORCA (40 cm radius)

When ORCA must maintain a slightly larger collision radius around the robots, the deadlock scenarios become increasingly apparent.