Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Recurso digital |
| Language: | |
| Published: |
Zenodo
2025
|
| Online Access: | https://doi.org/10.5281/zenodo.15032859 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866902338293202944 |
|---|---|
| author | Michalopoulos, Antonios Periklis Paliodimos, Efstratios Papadopoulos, Fotios Nikolaou, Grigoris Patrikakis, Charalampos Mitilineos, Stylianos |
| author_facet | Michalopoulos, Antonios Periklis Paliodimos, Efstratios Papadopoulos, Fotios Nikolaou, Grigoris Patrikakis, Charalampos Mitilineos, Stylianos |
| contents | <div> <h1>Project Description </h1> </div> <div> <p>During our research at the University of West Attica (UniWA), we addressed the problem of victim detection through walls in Search and Rescue (SAR) operations. To this end, we collected a dataset of radar signal recordings in a controlled classroom environment using a robotic system equipped with an ultra-wideband (UWB) pulsed radar. The collected data were then processed using various machine learning algorithms to detect a person behind a wall and to estimate its distance to the radar sensor. </p> </div> <div> <p>The experimental setup and the dataset acquisition methodology are thoroughly discussed in [1].</p> </div> <div> <div> <h1>Dataset Description </h1> </div> <div> <p>The dataset consists of 66 sessions of radar data corresponding to human presence. More specifically, data with six (6) victim actors were collected ("subjects"), across eleven (11) different, predefined robot positions, according to the setup described in [1]. Furthermore, 11 sessions of radar data were collected across the same robot positions, for human absence. Approximately 198 minutes of recorded human presence and 165 minutes of recorded absence are available, with a refresh rate of about 19 (radar) frames per second.</p> </div> <div> <div> <h1>Dataset Contents </h1> </div> <div> <div>Each subject folder is assigned an individual identifier Ni, where i=1...6 corresponds to subject number and N0 corresponds to subject absence. </div> <div>Within each subject folder, there are subfolders corresponding to the predefined robot positions as in [1], the subjects' orientation and distance to the wall. Within each subfolder, a CSV file stores a matrix of metadata and acquired radar data, according to the radar session data structure described in [1].</div> <div> <h1>Model Architecture</h1> </div> <div> <div>The explanation of the model architecture is presented in the file "CNN_Model.zip"</div> <div> </div> </div> </div> <div> <p>[1] Michalopoulos, A.-P et al, "Victim Detection Using a Robot-Mounted UWB-Radar Platform", submitted in the 14th International Conference on Modern Circuits and Systems Technologies (MOCAST 2025)</p> </div> </div> </div> |
| format | Recurso digital |
| id | zenodo_https___doi_org_10_5281_zenodo_15032859 |
| institution | Zenodo |
| language | |
| publishDate | 2025 |
| publisher | Zenodo |
| record_format | zenodo |
| spellingShingle | Victim Detection Using a Robot-Mounted UWB-Radar Platform Michalopoulos, Antonios Periklis Paliodimos, Efstratios Papadopoulos, Fotios Nikolaou, Grigoris Patrikakis, Charalampos Mitilineos, Stylianos <div> <h1>Project Description </h1> </div> <div> <p>During our research at the University of West Attica (UniWA), we addressed the problem of victim detection through walls in Search and Rescue (SAR) operations. To this end, we collected a dataset of radar signal recordings in a controlled classroom environment using a robotic system equipped with an ultra-wideband (UWB) pulsed radar. The collected data were then processed using various machine learning algorithms to detect a person behind a wall and to estimate its distance to the radar sensor. </p> </div> <div> <p>The experimental setup and the dataset acquisition methodology are thoroughly discussed in [1].</p> </div> <div> <div> <h1>Dataset Description </h1> </div> <div> <p>The dataset consists of 66 sessions of radar data corresponding to human presence. More specifically, data with six (6) victim actors were collected ("subjects"), across eleven (11) different, predefined robot positions, according to the setup described in [1]. Furthermore, 11 sessions of radar data were collected across the same robot positions, for human absence. Approximately 198 minutes of recorded human presence and 165 minutes of recorded absence are available, with a refresh rate of about 19 (radar) frames per second.</p> </div> <div> <div> <h1>Dataset Contents </h1> </div> <div> <div>Each subject folder is assigned an individual identifier Ni, where i=1...6 corresponds to subject number and N0 corresponds to subject absence. </div> <div>Within each subject folder, there are subfolders corresponding to the predefined robot positions as in [1], the subjects' orientation and distance to the wall. Within each subfolder, a CSV file stores a matrix of metadata and acquired radar data, according to the radar session data structure described in [1].</div> <div> <h1>Model Architecture</h1> </div> <div> <div>The explanation of the model architecture is presented in the file "CNN_Model.zip"</div> <div> </div> </div> </div> <div> <p>[1] Michalopoulos, A.-P et al, "Victim Detection Using a Robot-Mounted UWB-Radar Platform", submitted in the 14th International Conference on Modern Circuits and Systems Technologies (MOCAST 2025)</p> </div> </div> </div> |
| title | Victim Detection Using a Robot-Mounted UWB-Radar Platform |
| url | https://doi.org/10.5281/zenodo.15032859 |