Urban Remote Sensing. Группа авторов

users can have more direct control of the geographic extent of an area of interest (AOI), the spatial resolution of the image data, and the temporal resolution of the datasets by creating predefined flight missions. These variables can be influenced by adjusting the flight height and temporal intervals between multiple flight missions, respectively (Singh and Frazier, 2018). In this way, researchers can collect data only in the AOI with the desired resolution that can capture the variation of a phenomenon without getting needlessly too detailed (or needlessly taking too long). Many efforts have been made to improve urban mapping using active remote sensing through platforms mounted with high accuracy sensors, such as radar or light detection and ranging (LiDAR) sensors (Tison et al., 2004; Gonzalez‐Aguilera et al., 2012; Ban et al., 2015; Wurm et al., 2017). Due to advancements in photogrammetry, both spectral information and elevational information can be derived from UAS images. By mounting different sensors or cameras, UAS can also capture multi‐sourced information including hyperspectral information, thermal information, and laser scanning images in the same flight period, which can greatly very valuable for urban remote sensing. In addition to still imagery, UAS is also capable of recording videos, which can provide important geographic and environmental information of an AOI.

      This chapter discusses both the opportunities and challenges of UAS in urban applications. It is organized into five sections covering the advantages of UAS in urban remote sensing, common UAS models and camera types, UAS data collection and data processing, urban applications using UAS, major challenges and possible solutions, and conclusion and prospects.

3.2 COMMON UAS MODELS AND SENSORS

      3.2.1 COMMON MODELS

      In recent years, there has been dramatic technological development in the UAS models available for remote sensing applications. In addition to a notable increase in publications over the last decade regarding high‐accuracy remote sensing applications with UAS for aerial photogrammetry and three‐dimensional (3D) modeling (e.g. Remondino et al., 2011; Colomina and Molina, 2014; Toth et al., 2015; Agüera‐Vega et al., 2017; Erenoglu et al., 2018), there has also been an increase in the development of unique UAS platform designs for specific environmental settings (such as urban versus rural) (Chauhan, 2019; Yao et al., 2019). As more industries and disciplines are adopting the technology for their own purposes, more unique and specialized UAS models are being developed to meet the needs. Industries that operate in urban environments are developing UAS that can meet their unique environmental challenges, such as operations over tall buildings, power lines, high traffic, and groups of people, whereas other industries that operate in more rural settings might be focusing on other factors, such as covering large areas efficiently, flight length, and accessibility to the site of analysis. Although there are a wide variety of highly specialized UAS platforms available, most of them can be generally classified into either fixed‐wing or multi‐rotor systems (Saeed et al., 2018).

Fixed‐wing UAS are UA platforms that are configured like traditional airplanes with an airfoil (wings) and typically a single propeller. The forward airspeed of the UAS combined with the airfoil generates lift that enables the UAS to gain altitude. The use of surface controls on the airfoil provides fixed‐wing UAS with the ability to change the three dimensions of movement (i.e. pitch, yaw, and roll). These UAS types can fly much longer than multi‐rotor platforms. However, they also require larger areas to take‐off and land. Their extended flight times are generally a result of their lightweight airframe design, low‐energy cost (powering fewer propellers than a multi‐rotor), and manufacturer’s attention to the platform’s aerodynamics (Panagiotou and Yakinthos, 2020). These factors combine to yield greater gliding capacity for the UAS, which can significantly increase operational flight time as compared to multi‐rotor designs (Shan et al., 2017). Launching procedures for fixed‐wing UAS tend to be more user involved and require the use of a clearing to serve as a runway for take‐off and landing. Some fixed‐wing models use a launching method that requires an individual to carefully throw the UAS into the air while it is powered on, whereas other models utilize the ground and take‐off like traditional manned aircraft and therefore require a lengthy runway space to take‐off from. This is currently a well‐recognized challenge to the use of a fixed‐wing UAS in some environments with high amounts of obstacles present, such as in an urban setting or a heavily forested area, and there is ongoing research to address this challenge by making use of high‐accuracy GPS and vision sensors to improve fixed‐wing landings (Ding et al., 2018; Jantawong and Deelertpaiboon, 2018; Li et al., 2019; Lin et al., 2020). Due to the aforementioned reasons, fixed‐wing UAS are generally more suitable for agricultural or rural projects, like monitoring crop health (Shafian et al., 2018; Ziliani et al., 2018; Iizuka et al., 2019), mapping forests (Fraser and Congalton, 2018; Jayathunga et al., 2018), or analyzing land‐use/land‐cover (Hassan et al., 2011; Tsouros et al., 2019).

      Multi‐rotor UAS, on the other hand, are configured with multiple propellers in a symmetrical distribution around a central hub that allows positioning very precisely while the platform is in the air (Sámano et al., 2013). As Nascimento and Saska (2019) noted, there has been significant technological growth in multi‐rotor platforms over the last two decades. More specifically, the core components necessary for multi‐rotor designs to function in an efficient manner have improved in their technological capabilities, which has encouraged the large growth of the commercial UAS market. This technological growth can be seen in the surge of relatively low‐cost, yet high‐quality multi‐rotor platforms becoming more widely available for both research and commercial purposes since 2010 (Norouzi Ghazbi et al., 2016; Yao et al., 2019). Since a multi‐rotor UAS has multiple propellers operating at the same time, the flight times tend to be much shorter in duration compared to fixed‐wing platforms. The multiple propellers are in constant motion which generates vertical lift for the UAS, thus enabling the platform to hover in place. The ability to hover in place is what arguably makes multi‐rotor platforms more user‐friendly for less experienced UAS operators, as well as allow operators to conduct flights in more crowded areas. The ability to change direction with multi‐rotor UAS is enabled by varying each individual propeller’s thrust and torque, all of which are controlled by an onboard autopilot system that assists the pilot in controlling the positioning of the UAS so the pilot does not have to control each individual propeller. Due to the user‐friendliness, the ability for precision positioning and to carry heavier payloads a multi‐rotor frequently becomes the preferred platform for urban applications (González‐Jorge et al., 2017; Singh and Frazier, 2018; Watkins et al., 2020).

      There are other UAS models beyond the above two categories, which are much less commonly used in aerial remote sensing. These models are hybridized platforms that maintain certain characteristics of both fixed‐wing and multi‐rotor models. Hybrid platforms typically maintain the aerodynamic design of a fixed‐wing platform (large wingspan, lightweight) but can perform vertical take‐off and landing (VTOL) operations, much like a multi‐rotor platform. The ability for VTOL operations enables these platforms to be used in environments that generally do not allow for fixed‐wing UAS to safely take‐off and land in. Although these hybrid platforms are still less common than their fixed‐wing and multi‐rotor counterparts, we have witnessed a surge in developing more commercially viable systems that can meet unique industry demands using these hybridized designs (Floro da Silva and Branco, 2013; Thamm et al., 2015; Aktas et al., 2016; Hu and Lanzon, 2018; Joshi et al., 2019), which suggests a trend of hybrid models being more common and potentially more viable for commercial purposes as well.

      3.2.2 CAMERAS AND SENSORS

During the early integration of UAS technology into remote sensing in the early 2000s, visiblespectrum cameras (RGB) were the most popular type of payload sensor carried by commercially available UAS (Pajares, 2015). With the expansion of UAS technology into diverse industries more recently, the scaled‐down versions of many airborne and space imaging instruments are becoming available for UAS platforms, including multispectral sensors, hyperspectral sensors, thermal cameras, and LiDAR sensors. Detailed discussions about the selection of models and parameters (e.g. focal length and pixel size) for UAS sensing payloads can be found elsewhere (e.g. Colomina and Molina, 2014; Pepe et al., 2018; Yao et al.,