Unmanned Aerial Vehicle-Based Hazardous Materials Response: Information-Theoretic Hazardous Source Search and Reconstruction

Active Sensing for Search and Tracking: A Review

“…location of static targets that have already been detected [22], [24], [37], [39], [41]; Active Tracking differs from Active Localization in that the targets are dynamic [53]. Furthermore, in a Search and Tracking task, Active Search precedes (or coexists with) Active Tracking [35], [36], [42].…”

“…We say that a target is active when it occasionally [40] or continuously [37] releases information on their presence [47] (e.g., through radio ID [72]) or on their position (e.g., GPS SAT dynamic A > 1 > 1 [22], [39], [60] [41], [61], [62] AL static A 1 1…”

“…Alternatively, active targets propagate information on their position via radio broadcasting [39] or by spreading specific materials in the surroundings (e.g. hazardous chemicals, fluids, or gaseous particles [22]). The presence of active targets is usually captured by robots equipped with ad-hoc sensing capabilities (i.e., radio receivers [69], microphone arrays [66], gas detectors [24]); once the target is perceived, its position is inferred through the extraction of specific features from the collected data (e.g., the signal strength in radio communication [73]) Active targets often carry electronic devices that send information to the APE system (e.g., portable WiFi transmitters [74]).…”

“…In addition, relying on probabilistic strategies is necessary when dealing with large-scale search spaces; indeed, the inherent NP-hard complexity of the searching task [13], [82], [106]- [108] induces deterministic strategies to be computationally intractable for very large-scale environments [109], due to their polynomial [110] or exponential [111], [112] scalability with the size of the environment. Finally, compared to deterministic path planning, PTS strategies are more capable to adjust the platform behavior in response to new information [13], [52]; this higher adaptivity properties are proven to attain more accurate target position estimates and lower search times (i.e., higher efficiency) [22], [28]. Thus motivated, in this review we focus on PTS approaches; the reader may refer to [46], [49] for a broader discussion on all possible active search classes.…”

“…By leveraging the interplay among estimation, perception, and control, active sensing is at the nexus between automation and robotics research [7]. Fostered by the higher performance guarantees compared to passive approaches [8], [9] and enabling autonomous perception in robotics systems, active sensing is employed in Search and Rescue (SAR) missions [10]- [12], environmental mapping [13], [14] and exploration [15]- [17], wildlife monitoring [18]- [20], optimal sensor coverage [21], autonomous source term estimation (STE) [22]- [28], object recognition, classification and manipulation [9], [29], and collaborative mobile robotics [30]- [34].…”