Researchers released a dataset of 1,382 autonomous mobile robot jobs
The new MoRoOp dataset offers detailed logs of robot operations, including navigation errors and charging events in laboratory settings.
A research team has published MoRoOp, a new dataset tracking the behavior of autonomous mobile robots, or AMRs, as they handle warehouse-style tasks. The data covers nine full-day shifts where robots managed supply runs, filled empty boxes, and handled their own charging needs.
This release provides a deep look into how these machines actually perform under real-world conditions. Instead of showing only perfect runs, the data includes instances where robots faced navigation hurdles or failed to complete their tasks.
This dataset links job specifications, the operations constituting each job, dispatch events documenting operation state transitions and outcomes.
Key contents of the collection include:
- 1,382 individual jobs.
- 4,815 specific operational steps.
- 19,352 events marking state transitions.
- 140,386 sensor readings, including speed, position, and battery status.
Engineers can use this information to train agents to predict future movements or to detect if a machine is about to run into trouble. It also allows developers to build better simulations that mirror the messy reality of a busy workspace.
Imagine trying to teach a machine to navigate a hallway without having a map of every time it tripped over a loose cable. This dataset captures exactly that kind of friction. Have you ever wondered how many tiny decisions a robot makes just to move a box from one side of a room to another?
Because the data reflects a specific laboratory setup, it does not cover large-scale factory floors or outdoor environments. The researchers also noted that some operations were intentionally recorded as unsuccessful, meaning this is a snapshot of specific lab constraints rather than a universal model for all robotics.
Source: arXiv cs.RO