[Paper Review] RAWSim-O: A Simulation Framework for Robotic Mobile Fulfillment Systems

Marius Merschformann, Lin Xie+1 more|Multilingual Matters (Channel View Publications)|Oct 12, 2017

Advanced Manufacturing and Logistics Optimization4 references9 citations

TL;DR

RAWSim-O is a simulation framework for Robotic Mobile Fulfillment Systems (RMFS) that enables detailed, agent-based, event-driven modeling of robot movement, task allocation, and path planning in multi-floor warehouse environments. It integrates real-world robot prototypes (e.g., iRobot Create 2) with simulated systems to validate decision strategies, offering a scalable, open-source tool for optimizing RMFS performance through realistic kinematic modeling and hybrid simulation-real-world testing.

ABSTRACT

This paper deals with a new type of warehousing system, Robotic Mobile Fulfillment Systems (RMFS). In such systems, robots are sent to carry storage units, so-called "pods", from the inventory and bring them to human operators working at stations. At the stations, the items are picked according to customers' orders. There exist new decision problems in such systems, for example, the reallocation of pods after their visits at work stations or the selection of pods to fulfill orders. In order to analyze decision strategies for these decision problems and relations between them, we develop a simulation framework called "RAWSim-O" in this paper. Moreover, we show a real-world application of our simulation framework by integrating simple robot prototypes based on vacuum cleaning robots.

Motivation & Objective

• To address the lack of comprehensive simulation tools for Robotic Mobile Fulfillment Systems (RMFS) that model real-time decision problems such as robot task assignment, path planning, and pod allocation.
• To extend existing simulation frameworks like 'Alphabet Soup' by supporting multi-floor layouts connected by elevators and realistic robot kinematics, including turning time and acceleration/deceleration.
• To enable hybrid simulation-real-world validation by integrating low-cost robot prototypes (e.g., iRobot Create 2) with the simulation environment for practical testing of control and path planning algorithms.
• To support research on holistic system optimization by exposing synergies and side-effects of decision strategies, such as congestion from path planning choices.
• To provide an open-source, C#-based simulation framework compatible with high-performance clusters, facilitating reproducible research and algorithm development.

Proposed method

• The framework uses agent-based, event-driven simulation to model individual robots, pods, workstations, and elevators in a detailed, time-continuous manner.
• It incorporates realistic robot dynamics, including a 5.5-second turning time and acceleration/deceleration limits of ±0.5 m/s², to improve simulation fidelity.
• Path planning is performed using the MAPFWR solver to generate collision-free, deadlock-free, time-efficient trajectories, which are translated into discrete commands (go straight, turn left/right) for real robots.
• Real-world robot integration is achieved via iRobot Create 2 platforms equipped with ASUS Eee PCs, webcams for line-following, RFID tag readers for waypoint detection, and WiFi communication.
• A PID controller with OpenCV-based image processing is used for line-following, where binary filtering and morphological operations reduce noise and enable accurate centerline tracking.
• Synchronization between simulated and real robot movements is maintained by logging RFID waypoint data and aligning it with expected simulation states to minimize error accumulation.

Figure 1 : The central process of an RMFS (see [ 7 ] )

Experimental results

Research questions

• RQ1 How do real-time decision strategies for robot task assignment, path planning, and pod allocation interact and affect overall system efficiency in multi-floor RMFS?
• RQ2 To what extent can realistic robot kinematics, including turning time and acceleration constraints, be modeled in a simulation framework to improve real-world applicability?
• RQ3 Can a hybrid simulation-real-world setup using low-cost robot prototypes (e.g., iRobot Create 2) effectively validate path planning and control algorithms for RMFS?
• RQ4 What are the performance trade-offs and side-effects—such as congestion or bottlenecks—when optimizing individual components like robot routing or station workflow?
• RQ5 How can a simulation framework support the development and evaluation of scalable, multi-floor warehouse control strategies, especially those involving elevators as potential bottlenecks?

Key findings

• RAWSim-O successfully models complex RMFS dynamics, including multi-floor layouts with elevators, enabling analysis of vertical space utilization and elevator-related bottlenecks.
• The integration of real iRobot Create 2 robots with the simulation framework enables validation of path planning and control algorithms in a noisy, real-world environment with synchronized movement tracking.
• The use of realistic robot kinematics—such as a 5.5-second turning time and ±0.5 m/s² acceleration—improves simulation fidelity and reveals practical constraints not captured in idealized models.
• The hybrid simulation-real-world setup demonstrated that PID-based line-following with OpenCV image processing enables stable robot navigation, with error accumulation mitigated through state synchronization.
• The framework enables detection of system-level side effects, such as congestion from suboptimal path planning, which are critical for holistic performance assessment.
• The open-source release of RAWSim-O on GitHub, including controllers and algorithms, supports reproducible research and accelerates innovation in RMFS optimization.

(a) Overview of the simulation process.