January 20 2026

The paper A Comparative Study of Point-Based Deep Learning Techniques for Semantic Classification in Search and Rescue Arenas (2021), uses the ESOGU RAMPS dataset simulated from the NIST arena. The introduction gives a nice summary of the latest development in the Virtual Robotics competition. On this dataset PointNet, PointNet++, DGCNN, PointCNN, Point2Sequence, PointConv, and Shellnet point-based deep learning architectures were tested.
The dataset can be found Eskisehir Osmangazi University AI Robotics lab.
The dataset is described in A PointNet Application for Semantic Classification of Ramps in Search and Rescue Arenas (2019)
The dataset is created by augmenting the AI robotics lab with ramps with the NIST arena designer from the Hector team:
From the Hector 2012 TDP its clear that it is made by Johannes Simon. His bachelor thesis is called Visualizing Spatio-Temporal Data for Urban Management, but couldn't be found.
The MRESim paper is cited 13x
Adaptive Strategy for Environment Exploration in Search and Rescue Missions by Autonomous Robot is cited 3x
- Interesting is SAR with a Spot robot: Multi-floor danger and responsiveness assessment with autonomous legged robots in catastrophic scenarios.
Recent agent competition publication:
- Designing a Rescue Strategy Emphasizing Distributed Control in RoboCupRescue Simulation. cited by one: Information-Driven Team Collaboration in RoboCup Rescue (Dec. 2025). That last publication uses the agent simulator, but in quite a basic configuration:
Should also check and update Agent Simulation publications and symposium publications.
This publication cites several robocup rescue papers, but isn't itself one.

January 19, 2026

Found DARPA Subterranean Challenge: Multi-robotic Exploration of Underground Environments, and although the conference is on simulation, simulation is only mentioned as 'To address the importance of the software solutions, DARPA also organized a “virtual track” where the teams provide software solutions only, and the contest occurs in realistic simulations of the deployment sites. As expected, most teams regularly test their software in simulation as part of the development process, as simulation of the systems is much less tedious compared to real-world testing. Moreover, simulation allows us to test the individual software modules in isolation as well as in a holistic manner'
I should check if there is a survey article on the agent competition after "A survey and analysis of task allocation algorithms in multi-robot systems” (Cited 74x). Check:
- Optimization techniques for Multi-Robot Task Allocation problems: Review on the state-of-the-art (147x cited)
- Drone-surveillance for search and rescue in natural disaster (355x cited)
- A heuristic distributed task allocation method for multivehicle multitask problems and its application to search and rescue scenario (315x cited - no RoboCup link)
- Constraint optimization coordination architecture for search and rescue robotics (80x cited - no RoboCup link)

January 13, 2026

Looking further for citations of 16 Years of RoboCup Rescue (34x/49x), selecting the simulation oriented.
Also looked at the citations in Dimensions.
- The 2022 overview paper Rescue Robots for the Urban Earthquake Environment only indicates that more realistic simulations are needed.