Communication-Constrained Multi-Robot Exploration with Intermittent Rendezvous

TL;DR

The paper tackles multi-robot exploration under strict communication constraints by formulating the problem as a DEC-POMDP and introducing an intermittent rendezvous strategy. A rendezvous plan is automatically generated by mapping the task to a Job Shop Scheduling Problem (JSSP) and solving it with a Genetic Algorithm, yielding matrices $K$ (agreements) and $W$ (exploration steps) that encode when and with whom robots meet. Each robot operates a decentralized policy $\\pi_i$ that alternates between exploring frontiers and fulfilling rendezvous agreements, using a greedy joint-action heuristic within sub-teams to maximize DEC-POMDP rewards while sharing maps opportunistically. Simulation and Gazebo/ROS experiments show the approach can achieve faster exploration and higher rewards than base-station or relay-network baselines, validating the practicality of intermittent connectivity for scalable multi-robot exploration.

Abstract

Communication constraints can significantly impact robots' ability to share information, coordinate their movements, and synchronize their actions, thus limiting coordination in Multi-Robot Exploration (MRE) applications. In this work, we address these challenges by modeling the MRE application as a DEC-POMDP and designing a joint policy that follows a rendezvous plan. This policy allows robots to explore unknown environments while intermittently sharing maps opportunistically or at rendezvous locations without being constrained by joint path optimizations. To generate the rendezvous plan, robots represent the MRE task as an instance of the Job Shop Scheduling Problem (JSSP) and minimize JSSP metrics. They aim to reduce waiting times and increase connectivity, which correlates to the DEC-POMDP rewards and time to complete the task. Our simulation results suggest that our method is more efficient than using relays or maintaining intermittent communication with a base station, being a suitable approach for Multi-Robot Exploration. We developed a proof-of-concept using the Robot Operating System (ROS) that is available at: https://github.com/multirobotplayground/ROS-Noetic-Multi-robot-Sandbox.

Figures (9)

• Figure 1: Robots meeting at rendezvous locations spread over a construction site. The R1, R2, and R3 are hypothetical rendezvous locations. Colored arrows indicate the rendezvous locations. Orange circles are places to explore and dashed lines represent the paths followed by robots.
• Figure 2: Diagram for robot $i$ controller integrated with the rendezvous plan generator and the robot's components. The purple components generate a rendezvous plan through a JSSP, decide when and with whom robots should meet, and update the rendezvous locations. Differently, the light orange component refers to the exploration task and it makes robots explore unknown places. The robot's components are used for deployment and help assemble the robot's state and execute actions.
• Figure 3: Hypothetical agreements between $3$ robots (red, green, blue). Robot $1$ participates in the $1^{st}$, $3^{rd}$, and $4^{th}$ agreements, while robot $2$ participates in the $1^{st}$, $2^{nd}$, and $4^{th}$ ones. In the $1^{st}$ agreement, robots $1$ and $2$ should explore for $50$ time steps before going to a meeting location to share maps.
• Figure 4: Heuristic to pick new rendezvous locations when robots have common information during synchronization for $3$ sub-teams from the example from Fig. \ref{['fig:agreements']} in a construction site. The orange circles are frontiers, and the blue circles are the center of masses obtained from the KNN algorithm. The center of masses $c_1$ and $c_3$ were chosen to the $1^{st}$ and $2^{nd}$ sub-teams. Differently, $c_2$ was chosen to the $3^{rd}$ sub-team by maximizing the distance $\sqrt{(c_1 - c_2)^2 + (c_2 -c_3)^2}$.
• Figure 5: Samples from our benchmark and simulator. (From left to right) Sections of London, Dubai, Washington DC, Beijing, Rio de Janeiro, construction site, and Dallas International Airport.