Autonomous mobile robot operations logistics: a dataset of jobs, dispatch events and robot states

Jan-Felix Klein; Yongkuk Jeong

cs.ROOct 1, 2026

Autonomous mobile robot operations logistics: a dataset of jobs, dispatch events and robot states

Authors: Jan-Felix Klein, Yongkuk Jeong

Organizations: Department of Production Engineering, KTH Royal Institute of Technology, Stockholm, Sweden · Digital Futures, KTH Royal Institute of Technology, Stockholm, Sweden

Abstract

Autonomous mobile robots (AMRs) increasingly perform material transport in production logistics, where their operation is governed by job generation, dispatching and robot control. We present MoRoOp, a dataset of AMR operations recorded in a laboratory kit preparation and supply scenario over nine eight-hour shifts. During each shift, an AMR executed stochastically generated kit supply, empty-box refill and charging jobs. The dataset links job specifications, the operations constituting each job, dispatch events documenting operation state transitions and outcomes, and robot-state observations comprising position, orientation, velocity, per-wheel state of charge and diagnostics. It contains 1,382 jobs, 4,815 operations, 19,352 dispatch events and 140,386 robot-state observations together with the kit specifications used during job generation. The dataset was recorded in an operating laboratory environment, and technically valid observations of delays, obstructed navigation and unsuccessful operations were retained. Both raw and cleaned robot-state tables are provided. Documented reuse directions include the evaluation of AI agents on operational decision records, disturbance detection, operation prediction, data-driven simulation and event-log analysis.

Figures & tables

Figure 1: An overview of the production logistics scenario at the IPU lab.

Figure 2: Job generation procedure.

Symbol	Type	Description	Value or distribution
$i$	Index	kit supply job index	$i\in\mathbb{N}$
$t$	Time variable	Current time relative to the beginning of the shift	$t\in[0,T_{S}]$
$t_{g,i}$	Event time	Generation time of kit supply job $i$	$t_{g,i}\in[0,T_{S}]$
$T_{arr,i}$	CRV	Inter-arrival time preceding kit supply job $i$	$\operatorname{Tri}(225,250,275)$
$K_{i}$	DRV	Kit sampled for kit supply job $i$	$K_{i}\in\mathcal{K}=\{K_{001},\ldots,K_{030}\}$
$c(K_{i})$	Derived variable	Number of unitized component boxes required for kit $K_{i}$	$c(K_{i})\in\{2,3\}$

Table 1: Variables and parameters used in the job generator. All event times, durations, and processing times are specified in seconds.

\operatorname{Tri}(a,m,b)

denotes a triangular distribution with minimum

a

, mode

m

, and maximum

b

. CRV denotes a continuous random variable; DRV denotes a discrete random variable, FP a fixed parameter.

Figure 4: The folder structure of the MoRoOp dataset.

Field	Data type	Description
kit_id	string	Unique identifier of the kit to which the brick variant belongs. Multiple rows may share the same kit identifier.
color	string	Color of the brick, either blue , red , yellow or green .
size	string	Size of the brick variant, either small or large .
quantity	int	Number of bricks of the specified type required by the kit.

Table 2: Fields contained in the kits.parquet table.

Field	Data type	Description
id	string	UUIDv4 unique identifier for the job.
created_at	timestamp	Timestamp at which the job record was created.
payload	string	JSON-encoded job specification, including job type and kit identifier.
operations	string	JSON-encoded ordered list of operation identifiers belonging to the job.
state	string	Recorded lifecycle state of the job.
outcome	string	Recorded execution outcome of the job.

Table 3: Fields contained in the *_jobs.parquet table.

Field	Data type	Description
id	string	UUIDv4 unique identifier for the operation.
created_at	timestamp	Timestamp at which the operation record was created.
payload	string	JSON-encoded operation specification, including applicable operation-specific parameters.
priority	int	Numeric priority assigned to the operation.
job_id	string	Identifier of the job to which the operation belongs.
dependencies	string	JSON-encoded list of dependencies linking the operation to required preceding operations.

Table 4: Fields contained in the *_operations.parquet table.

Field	Data type	Description
id	string	UUIDv4 unique identifier for the dispatch event.
created_at	timestamp	Timestamp at which the event was created.
event_type	string	Recorded event type.
operation_id	string	Identifier of the operation associated with the event, where applicable.
changes	string	JSON-encoded description of the operation attributes changed by the event.
meta_data	string	JSON-encoded supplementary information associated with the event, where applicable.

Table 5: Fields contained in the *_dispatch_events.parquet table.

Field	Data type	Description
id	int	Sequential identifier assigned to observation.
created_at	timestamp	Timestamp at which the robot state was recorded.
operating_mode	string	Robot operating mode at the time of observation.
state	string	Robot state at the time of observation.
stuck_reason	string	Reported reason the robot was stuck, if any.
pos_x	double	Robot x-coordinate in the map reference frame.

Table 6: Data dictionary for the robot_state table.

Panel of Figure 5 — Figure 5: Representative operation sequence comprising the supply of kit K014 and an empty-box refill job during the 2026_07_27_evening shift.

Panel of Figure 6 — Figure 6: Representative robot-state data from the 2026_07_27_evening shift.

shift	dispatch events	jobs	operations	robot state (raw)	robot state (cleaned)
2026_07_27_morning	2,149	154	536	15,549	15,194
2026_07_27_evening	2,145	154	535	15,389	15,038
2026_07_28_morning	2,173	156	540	15,912	15,553
2026_07_28_evening	2,159	155	538	15,566	15,252
2026_07_29_morning	2,170	155	541	15,546	15,182
2026_07_29_evening	2,167	155	539	15,491	15,168

Table 7: Number of records in each dataset table by shift.

Mobile Robotics

Mobile robotics studies how ground, aerial, and aquatic robots perceive, navigate, and act in changing real-world environments.

See all 135 papers

Scalable Trajectory Generation for Whole-Body Mobile Manipulation
Apr 14, 2026Yida Niu, Xinhai Chang, Xin Liu +2Robot Motion Generation Mobile Manipulation
Modeling Robotics Dataset Construction as an Artifact-Based Build Process
May 29, 2026Leon Pohl, Lukas Beer, George Sebastian +1Robotics Automated Software Engineering
Institute for Autonomous Driving, University of the Bundeswehr Munich, 85579 Neubiberg, Germany
OHRID-Retail: An Open Multimodal Dataset of Human Activity in Retail Environments
Sep 16, 2026Xiangrui Wang, Yuetong Wu, Jalen Beeman +6Human-Robot Interaction
Department of Industrial and Systems Engineering, University of Florida, Gainesville, FL, USA. · Lane Department of Computer Science and Electrical Engineering, West Virginia University, Morgantown, WV, USA · Department of Mechanical, Materials and Aerospace Engineering, West Virginia University, Morgantown, WV, USA. +1

Apr 14, 2026J/K move · Enter open · S save

Scalable Trajectory Generation for Whole-Body Mobile Manipulation

Yida Niu, Xinhai Chang, Xin Liu, Ziyuan Jiao+1

Robot Motion Generation Mobile Manipulation Whole-Body Robotic Manipulation Synthetic Data Generation for Robotics Trajectory Optimization

Open paper AskPDF arXiv

Robots deployed in unstructured environments must coordinate whole-body motion -- simultaneously moving a mobile base and arm -- to interact with the physical world. This coupled mobility and dexterity yields a state space that grows combinatorially with scene and object diversity, demanding datasets far larger than those sufficient for fixed-base manipulation. Yet existing acquisition methods, including teleoperation and planning, are either labor-intensive or computationally prohibitive at scale. The core bottleneck is the lack of a scalable pipeline for generating large-scale, physically valid, coordinated trajectory data across diverse embodiments and environments. Here we introduce AutoMoMa, a GPU-accelerated framework that unifies AKR modeling, which consolidates base, arm, and object kinematics into a single chain, with parallelized trajectory optimization. AutoMoMa achieves 5,000 episodes per GPU-hour (over

80\times

faster than CPU-based baselines), producing a dataset of over 500k physically valid trajectories spanning 330 scenes, diverse articulated objects, and multiple robot embodiments. Prior datasets were forced to compromise on scale, diversity, or kinematic fidelity; AutoMoMa addresses all three simultaneously. Training downstream IL policies further reveals that even a single articulated-object task requires tens of thousands of demonstrations for SOTA methods to reach

\approx 80\%

success, confirming that data scarcity -- not algorithmic limitations -- has been the binding constraint. AutoMoMa thus bridges high-performance planning and reliable IL-based control, providing the infrastructure previously missing for coordinated mobile manipulation research. By making large-scale, kinematically valid training data practical, AutoMoMa showcases generalizable whole-body robot policies capable of operating in the diverse, unstructured settings of the real world.

Abstract

Mobile Robotics

Explore similar work

Scalable Trajectory Generation for Whole-Body Mobile Manipulation

Modeling Robotics Dataset Construction as an Artifact-Based Build Process

OHRID-Retail: An Open Multimodal Dataset of Human Activity in Retail Environments