Radiation search operations using scene understanding with autonomous UAV and UGV

Abstract

Autonomously searching for hazardous radiation sources requires the ability of the aerial and ground systems to understand the scene they are scouting. In this paper, we present systems, algorithms, and experiments to perform radiation search using unmanned aerial vehicles (UAV) and unmanned ground vehicles (UGV) by employing semantic scene segmentation. The aerial data is used to identify radiological points of interest, generate an orthophoto along with a digital elevation model (DEM) of the scene, and perform semantic segmentation to assign a category (e.g. road, grass) to each pixel in the orthophoto. We perform semantic segmentation by training a model on a dataset of images we collected and annotated, using the model to perform inference on images of the test area unseen to the model, and then refining the results with the DEM to better reason about category predictions at each pixel. We then use all of these outputs to plan a path for a UGV carrying a LiDAR to map the environment and avoid obstacles not present during the flight, and a radiation detector to collect more precise radiation measurements from the ground. Results of the analysis for each scenario tested favorably. We also note that our approach is general and has the potential to work for a variety of different sensing tasks.

Online Aerial Terrain Mapping for Ground Robot Navigation

“…The remaining area is assigned to the grass class. This classifier used hand-tuned thresholds and was not particularly robust to variability in the conditions, it only serves as a placeholder for more sophisticated terrain classification algorithms, such as the algorithm described in [ 1 ], which will be integrated at a later date.…”

“…Planners typically minimize path length which does not necessarily minimize energy consumption. In radiation search operations using scene understanding with autonomous unmanned aerial vehicles (UAVs) and unmanned ground vehicles (UGVs) in [ 1 ], the authors show the difference in the power consumed by a ground vehicle traversing grass and pavement. This work utilizes overhead imagery to derive terrain classes, infers traversibility costs from these terrain classes, generates an energy-efficient trajectory through this cost map, and then commands the ground vehicle to follow this trajectory.…”

The Challenge of Preparing Teams for the European Robotics League: Emergency

“…humanoid) robot as such excellent multipurpose robot might be very difficult to design and/or prohibitively expensive to purchase and maintain. In search and rescue robotics, UAVs and UGVs have been researched in collaborative scenarios for mapping, localization and task planning in multiple cases [5], [6], [8], [10].…”

Experiments in unmanned aerial vehicle/unmanned ground vehicle radiation search

“…Christie et al () validate an end‐to‐end multirobot system for the inspection of anomalous sources of radiation in hardware experiments. First a UAV performs a complete coverage flight of the environment simultaneously collecting radiation measurements as well as stereo imagery of the area.…”

“…While wavelet scattering transforms are useful for representing functions where there is some notion of scale (Bruna & Mallat, 2013), several authors of this paper have developed an alternative FST (Czaja & Li, 2017a,b) that is particularly well suited for spectral data. Christie et al (2016) validate an end-to-end multirobot system for the inspection of anomalous sources of radiation in hardware experiments. First a UAV performs a complete coverage flight of the environment simultaneously collecting radiation measurements as well as stereo imagery of the area.…”