Comparative Analysis of Priority-based Sequential and Thread-based Concurrent Task Execution Strategies for Multi-Robot Warehouse Automation

Abstract

A multi-robot system can improve time efficiency
in warehouse transportation; however, some applications require
predictability in addition to efficiency. The task execution strategy
selected for a multi-robot system significantly influences task
completion time, predictability, and system complexity. This
paper presents an empirical study comparing two task execution
strategies for a multi-robot warehouse platform: (1) a prioritybased
sequential approach, where a central controller runs a
single thread to advance each robot’s state individually, and (2) a
thread-based concurrent approach, where each robot is managed
by its own thread and coordinates through periodic checks of
shared pose data. Both were tested against a single-robot baseline.
Each robot is built with an ESP32 microcontroller, encoders for
motion feedback, and a servo-driven gripping mechanism. Robots
operate in a controlled platform that consists of drop zones and
loads. The whole system uses ArUco marker-based overhead
localization and a ROS 2 central controller communicating with
each robot over UDP/JSON. Across the repeated trials of both
task execution strategies on the system with varying load counts,
the thread-based concurrent execution strategy finished tasks
27–49% faster but had 39–78% more timing unpredictability,
compared to the priority-based sequential task execution strategy.