2.3.1 Autonomous Urban Search and Rescue
Since the turn of the century, mobile robots have become extremely useful resources in USAR due to their ability to search unknown areas that are hazardous or even inaccessible to rescue workers [Liu & Nejat, 2013]. There are many advantages of using robots in USAR: unlike rescue workers, they are not affected by stress or fatigue [Burke, Murphy, Coovert & Riddle]; robots can be made in large quantities whereas human rescue workers are not as readily available [Casper & Murphy, 2003]; damaged robots can be easily repaired or replaced whereas the loss of a human life has a much greater impact on society [Casper, Micire & Murphy, 2000]. Due to these obvious benefits, the use of mobile robots in USAR has become a significant research topic in the last decade [Liu & Nejat, 2013].
The development of rescue robots was motivated by two major disasters: the Kobe earthquake in Japan in 1995 and the Oklahoma City bombing in 1996 [Murphy, Tadokoro, Nardi, Jacoff, Fiorini, Choset & Erkmen, 2008]. Since the turn of the century, robots have been utilised in many USAR operations. The first known application of robots in USAR was in the aftermath of the World Trade Centre disaster on September 11th 2001 [Murphy, 2004]. Since then, mobile robots have participated in USAR operations of many other disasters, such as Hurricanes Katrina, Rita, and Wilma in the U.S.A. in 2005 [Murphy et al, 2008], the Haiti earthquake in 2010 [Guizzo, 2011], and the Tohoku earthquake and tsunami in Japan in 2011 [Guizzo, 2011].
One of the main challenges associated with autonomous USAR is that disaster sites are often highly cluttered, which makes it very difficult for robots to navigate the site and search for survivors without a human in the loop [Liu & Nejat, 2013]. However, this has its own problems, as humans may have difficulties in determining the true nature of the environment from remote visual feedback [Liu & Nejat, 2013], and this can result in robots becoming physically stuck [Casper & Murphy, 2003]. Therefore, one of the main aims of robotics research is to improve low-level
Chapter 2 Literature Review
11 autonomy, such as controlling robots for the purpose of navigation through rough terrain [Mourikis, Trawny, Roumeliotis, Helmick & Matthies, 2007; Okada, Nagatani, Yoshida, Tadokoro, Yoshida & Koyanagi, 2011], and also being able to map an USAR environment [Kurisu, Muroi, Yokokohji & Kuwahara, 2007; Zhang, Nejat, Guo & Huang, 2011]. There has also been a great deal of research in the development of semi-autonomous control [Wegner & Anderson, 2006; Doroodgar, Ficocelli, Mobedi & Nejat, 2010], which can provide a balance between teleoperation and low- level autonomy. This type of balance is very useful as it allows the operator to concentrate on higher-level tasks such as supervision of multiple robots and specifying the direction of travel [Liu & Nejat, 2013], and has thus paved the way for the development of single-human multi-robot systems. Such systems are more cost-effective than single-human single-robot systems [Liu & Nejat, 2013], and therefore, a great deal of research has been done on the development of such systems, with much emphasis being placed on teamwork among the robots themselves, and also between robots and humans [Sato, Matsuno, Yamasaki, Kamegawa, Shiroma & Igarashi, 2004; Luo, Espinosa, Pranantha & De Gloria, 2011]. The development of autonomous USAR systems in recent years has certainly proved to be very promising, but there are still more challenges that lie ahead, including the development of robotic systems that can transport trapped victims to safety [Yim & Laucharoen, 2011], and therefore, autonomous USAR promises to be a very exciting research area in the coming years.
While the contribution of this thesis is in the development of autonomous air-sea rescue rather than autonomous USAR, the methodologies are certainly transferrable to USAR. This thesis aims to determine whether optimisation techniques can be applied to autonomous air-sea rescue so that a search can be carried out in a structured and controlled manner, and extends on the work carried out by Worrall (2008) by introducing several hybrid algorithms.
2.3.2 Standard Maritime Search and Rescue Techniques
Currently, the standard procedures carried out in air-sea search and rescue operations can be found in the International Aeronautical and Maritime Search and Rescue (IAMSAR) manual [IAMSAR, 2008]. In particular, the standard search approaches are outlined in Volume II, Mission co- ordination. Some of the most common search patterns used are the Parallel Sweep Search, Sector Search, and Expanding Square Search.
The Parallel Sweep Search simply sweeps back and forward along the long sides of a rectangle, moving part of the way along the smaller side between each sweep. The search can be carried out with multiple vehicles by assigning each vehicle to separate sub-regions. This technique is often used when there is a large uncertainty in the target locations [IAMSAR, 2008], and is very effective at searching areas with uniform coverage. This technique is described in more detail in Chapter 5.
Chapter 2 Literature Review
12 The Sector Search is used to search a circular area about some point. The search starts at the centre and travels to the edge of the circle, then turns 120° starboard, then keeps on searching, turning 120° starboard every time it reaches the edge of the circle. Consequently, this particular search gives good coverage nearer the centre of the circle, and is very effective when target locations are known reasonably well and also when the search area is small [IAMSAR, 2008]. Like Parallel Sweep, this technique is described in more detail in Chapter 5.
The Expanding Square Search starts at the centre of a given region, and then travels around the centre point in a square pattern, with the length of the square expanding after every two sides, so that the search covers the area around the centre in a uniform manner. Like the Sector Search, this technique is often most effective when the target locations are known reasonably well [IAMSAR, 2008]. Again, this technique is described in more detail in Chapter 5.
There are many other search techniques, and various criteria for using each particular search method. As mentioned, more details on these search methods can be found in the IAMSAR manual [IAMSAR, 2008]. In this thesis, the three techniques described above provide a benchmark for more complex heuristic techniques in an air-sea search mission, where the heuristic techniques are developed from existing optimisation methods, which are not normally used in this context.