The Critical Objects for Response to Emergency (CORE) dataset was collected to facilitate research on robot perception and reasoning in search and rescue applications. The CORE dataset includes 23 objects in total: 100ft rope, bolt cutter, clear safety mask, electric couple, electric drill, electric grinder, emergency light, fire extinguisher, firemans axe, hard hat, heavy hammer, micro torch welder, mine communication amplifier, oxygen mask, pry bar, reciprocating saw, wrench, self-rescuer, leather work glove, steel toe boot, tag reader, walkie talkie, and welding mask. The objects are selected based on the discussion with our first responder collaborators in underground mine and urban search and rescue teams. An example 3D point cloud of each critical object category is illustrated in the following figure.


The objects in the CORE dataset are selected based on their utility and importance to rescuers and other first responders in search and rescue missions (especially in underground rescue operations). Some objects are critical in routine safety checking. For example, since RF tag readers indicate whether miners are in the underground mine, determining whether the device is functional or salvageable is necessary. In addition, we consider the objects that are essential to find the trace of victims in a diaster environment. For example, self-rescuers usually can provide a good indication of victims' locations, since miners usually release them in case of carbon monoxide gas. Also, detection and location of certain assets and tools is critical, especially in situations where resources are limited. Therefore, various tools (such as drills and grinders) are also added to the dataset.

Download and Citation

Download the CORE dataset at Version 1.0. The dataset provided here is for non-commercial research/educational use only.
We are now working on collecting a new version of the CORE dataset in real underground mine environments.

Please cite one of the following papers if you use this dataset. Detailed information of the CORE dataset and our approaches are also available in these papers.

Questions and Suggestions

Please contact Ahmed Ambarak: aambarak -AT- mymail -DOT- mines -DOT- edu