MRTA-Sim: A Modular Simulator for Multi-Robot Allocation, Planning, and Control in Open-World Environments

TL;DR

MRTA-Sim addresses the need for realistic, open-world MRTA evaluation by providing a modular, physics-based simulator that unifies task allocation, path planning, and control with deconfliction mechanisms. It couples an SMT-based or SPECLESS allocation solver with a road-networked waypoint framework, a Nav2-based planning service, and a CBF-QP controller that can operate in both decentralized and cluster-based centralized modes. Key contributions include the MRTA-Sim architecture, a suite of baseline components, Scenic integration for scenario generation, and scalability experiments demonstrating real-time feasibility for modest fleets. The platform enables end-to-end testing of MRTA approaches in long-running tasks and complex environments, facilitating comparisons and iterative development toward more robust multi-robot systems in real-world settings.

Abstract

This paper introduces MRTA-Sim, a Python/ROS2/Gazebo simulator for testing approaches to Multi-Robot Task Allocation (MRTA) problems on simulated robots in complex, indoor environments. Grid-based approaches to MRTA problems can be too restrictive for use in complex, dynamic environments such in warehouses, department stores, hospitals, etc. However, approaches that operate in free-space often operate at a layer of abstraction above the control and planning layers of a robot and make an assumption on approximate travel time between points of interest in the system. These abstractions can neglect the impact of the tight space and multi-agent interactions on the quality of the solution. Therefore, MRTA solutions should be tested with the navigation stacks of the robots in mind, taking into account robot planning, conflict avoidance between robots, and human interaction and avoidance. This tool connects the allocation output of MRTA solvers to individual robot planning using the NAV2 stack and local, centralized multi-robot deconfliction using Control Barrier Function-Quadrtic Programs (CBF-QPs), creating a platform closer to real-world operation for more comprehensive testing of these approaches. The simulation architecture is modular so that users can swap out methods at different levels of the stack. We show the use of our system with a Satisfiability Modulo Theories (SMT)-based approach to dynamic MRTA on a fleet of indoor delivery robots.

Figures (9)

• Figure 1: View of robots (center of the blue circles) moving in the simulation. Hospital beds, crowded rooms, humans, and tight doorways create difficult movement challenges for multi-agent, continuous operation.
• Figure 2: The Gazebo view is shown in the top row; the bottom row shows the Rviz visualization of paths, obstacle points, and agent positions and headings. From left to right: a) The yellow and pink agents follow the "right-hand" travel rule on the roads. The red agent is waiting for the higher priority pink agent to pass first due to the CBF cluster control b) A moving human is shown in Gazebo and Rviz. c) Due to CBF cluster control, the green agent has moved out of the way of the red agent that just arrived. d) The blue agent originally plans for the end of the queue (the right-most small dot) before it knows its queue position. Once it receives its position, it adjusts its path to plan towards the room. The red agent is not in the queue because it has completed its tasks.
• Figure 3: MRTA-Sim Architecture
• Figure 4: The bookshelf in these two screenshots has been added via a Scenic program. The distance away from the wall for each was automatically sampled from a distribution by Scenic.
• Figure 5: From left to right: All agents are stopped because they have no active tasks. Then the blue and green agents received tasks. The green agent waits for the higher priority blue agent to pass. Then, it replans towards its goal and moves to follow its path.