For safe and efficient collaborative environments, co-located robots must be able to visually estimate and track the movements of surrounding robots. Nonparametric belief propagation provides a probabilistic framework for autonomous systems to reason about uncertainty and multiple hypotheses at once, as well as leverage portions of the robot that are better observed. Due to its computational constraints on the sample size, belief propagation often causes the filter to converge to incorrect estimates, because the particles are ineffectively representing the true belief. Our work seeks to maintain particle diversity by re-initializing the particles from a variety of proposal distributions as needed. We extend promising work of adaptive particle reinvigoration from a single distribution in the particle filter domain, which introduced Counter-Hypothetical reasoning to independently estimate when the filter was in failure mode. Our proposed framework explores the usefulness of this reasoning within belief propagation to manage sampling from multiple distributions and propagating its information through the standard graphical model. We present preliminary qualitative results for this method on tracking 21 links on a humanoid robot, Digit. 

Visual Robot Pose Tracking through Counter-Hypothetical Nonparametric Belief Propagation

Decentralized control for multi-robot systems involves planning in complex, high-dimensional spaces. The planning problem is particularly challenging in the presence of potential collisions between robots and obstacles, and different sources of uncertainty such as inaccurate dynamic models and sensor noise. A multi-robot system can be represented as a graphical model, in which nodes represent individual robots and edges represent communication between robots. This representation enables the use of graphical inference algorithms for solving multi-robot control. In this short paper, we introduce Stein Variational Belief Propogation (SVBP), a novel algorithm for performing inference over the marginal distributions of nodes in a graph. We present simulation results which demonstrate that our method can represent complex, multi-modal distributions in localization and control tasks.

Stein Variational Belief Propagation for Decentralized Multi-Robot Control

In many robotics problems, there is a significant gain in collaborative information sharing between multiple robots, for exploration, search and rescue, tracking multiple targets, or mapping large environments.  
One of the key implicit assumptions when solving cooperative multi-robot problems is that all robots use the same (homogeneous) underlying algorithm. 
However, in practice, we want to allow collaboration between robots possessing different capabilities and that therefore must rely on heterogeneous algorithms.
We present a system architecture and the supporting theory, to enable collaboration in a decentralized network of robots, where each robot relies on different estimation algorithms.
To develop our approach, we focus on multi-robot simultaneous localization and mapping (SLAM) with multi-target tracking.
Our theoretical framework builds on our idea of exploiting the conditional independence structure inherent to many robotics applications to separate between each robot’s local inference (estimation) tasks and fuse only relevant parts of their non-equal, but overlapping probability density function (pdfs).
We present a new decentralized graph-based approach to the multi-robot SLAM and tracking problem.
We leverage factor graphs to split between different parts of the problem for efficient data sharing between robots in the network while enabling robots to use different local sparse landmark/dense/metric-semantic SLAM algorithms.

Toward Decentralized Heterogeneous Multi-Robot SLAM and Target Tracking

The autonomous operation of drone swarms cur-
rently poses a challenge due the limited bandwidth and reliability
of drone-to-drone or drone-to-controller communication. In this
paper we present a decentralised drone-to-drone communication
algorithm for synchronisation that uses a mesh network topology.
Current approaches use centralised control and communication
organised in a star topology that prevents drone-to-drone commu-
nication. This poses a single point of control an communication
failure and scalability problems. Drone-to-drone communication
organised as a mesh network is a viable option that could increase
swarm autonomy and resilience to communication failure. In this
paper we propose the Epoch Synchronisation with Pulse Coupled
Oscillators (ESPCO) distributed algorithm for decentralised
synchronisation. We design, implement, and demonstrate a Pulse
Coupled Oscillator based agreement protocol that is decentralised
and robust to individual drone failure.

Pulse Coupled Oscillators for Drone-to-Drone Synchronisation

We will present a demonstration of a Robot Web for Distributed Many-Device Localisation. 
Robot Web is a distributed network of robots or other devices that can collaborate to achieve global localisation by making measurements of each other through efficient ad-hoc peer-to-peer communication. The system utilises Gaussian Belief Propagation on the underlying nonlinear factor graph, which describes the probabilistic structure of all observations made by robots internally or of one another.
In this proposed demonstration, we will show Robot Web running on a multi-robot simulation with an interactive GUI.


Demo of a Robot Web for Distributed Many-Device Localisation

We introduce the swarming oscillator framework to robotics researchers and present exciting preliminary results related to distributed coupling. The full development of this study is a full research program that will consist of various phases across many papers and studies. One phase will enable us to refine distributed coupling swarming oscillator models so that we can achieve specific swarming behaviors. The next phase uses our distributed coupling models to design robot collective systems that realize the swarming behavior predicted through the swarming oscillator model. In the first phase, this work will demonstrate a major thrust for the swarming oscillator field and will have the potential to impact diverse fields ranging from swarm robotics, to mobile networking systems, to characterization of natural swarming phenomena.

Towards Self-Organizing Swarms Composed of Swarmalators with Distributed Couplings

Collaborative Simultaneous Localization and Mapping (CSLAM) is a critical capability for enabling multiple robots to operate in complex environments. Most CSLAM techniques rely on the transmission of low-level features for visual and LiDAR-based approaches, which are used for pose graph optimization. However, these low-level features can lead to incorrect loop closures, negatively impacting map generation. Recent approaches have proposed the use of high-level semantic information in the form of Hierarchical Semantic Graphs to improve the loop closure procedures and overall precision of SLAM algorithms. In this work, we present Multi S-Graphs, an S-graphs based distributed CSLAM algorithm that utilizes high-level semantic information for cooperative map generation while minimizing the amount of information exchanged between robots. Experimental results demonstrate the promising performance of the proposed algorithm in map generation tasks.

Multi S-graphs: A Collaborative Semantic SLAM architecture.

A swarm of robots is often required to create formations in space whilst avoiding obstacles in the environment. In order to do so a high degree of coordination is required between the robots to ensure smooth and safe paths. We formulate formation planning as performing inference on a factor graph using Gaussian Belief Propagation (GBP). This work is an outline of the real time and interactive demos we would like to show at the Distributed Graphs Workshop at ICRA 2023. We show some examples of the demos that will be displayed, and hope that this work encourages more research in this field.

Distributed Formation Planning for Robot Swarms

Multi-robot herding deals with the problem of steering a group of non-cooperative entities towards desired locations by means of a team of robots. Herding demands high levels of coordination due to the complex nonlinear dynamics of such evasive entities, and the wild environments preclude the use of central coordination units. These aspects lead to the need of distributed estimation and graph-based control protocols. In this work, we propose CIC (Coordinated Implicit Control), a distributed control protocol that is composed by two main blocks. The first is an Implicit Control module that is able to stabilize any continuous input-nonaffine system, tackling the complex nature of the herd dynamics. The second is an Extended Certifiable Optimal Distributed Kalman Filter that, by defining the multi-robot team as a distributed communication graph, allows each robot to rapidly recover the information necessary to compute the control input, while accounting for potential uncertainties. Different experiments validate the proposal. 

ICRA 2023 Workshop DGA Submissions