1 Introduction

Recent approaches in the optimization techniques are mostly based on natural phenomenon and behavioral observation. Genetic algorithms (GA) (Glover 1994) and evolutionary algorithms (EA) (Thomas 1996), generate solutions to optimization problems using techniques inspired by natural evolution, such as inheritance, mutation, selection and crossover. Harmony search (HS) (Geem et al. 2001) is inspired by the improvisation process of musicians. Intelligent water drops algorithm (IWD) (Duan et al. 2009) is a nature-inspired optimization algorithm, which is based on the natural flow of rivers and how they find almost optimal paths to their destinations. Gravitational search algorithm (GSA) (Rashedi et al. 2009) is constructed based on the law of gravity and the motion of mass interactions. On the other hand, techniques related to behavioral observation include the ant colony optimization (ACO) (Dorigo et al. 2006), which is a class of optimization algorithms modeled on the actions of an ant colony. Artificial immune systems (AIS) are computational systems inspired by the principles and processes of the vertebrate immune system. Particle swarm optimization (PSO) (Poli et al. 2007) is inspired by the social behavior of bird flocking or fish schooling. Photosynthetic algorithm (PA) (Murase 2000) utilizes the dark reaction rules governing the transfer of carbon molecules from one substance into another in the Calvin–Benson cycle and photorespiration. Galaxy-based search algorithm (GbSA) (Shah Hosseini 2011) imitates the spiral arm of spiral galaxies in searching for its surroundings. The list is expanding, so much so that generally, we can classify these algorithms into sub categories of its own. i.e. Mother Nature based systems, biological based etc.

In the earlier years, the creation of a robot is mostly based on the anatomy and the movement of human beings. They are generically cylindrical in shape, having static hands, wheels like a car and their purpose was to assist in the construction industry, manufacturing as well as the search and rescue operations. However, the development of bio-mimetics (nature inspired) in the early 1990s has “forced” researchers to begin winds of change. The era of animal-like anatomy copying began. Each shape of the animals has its own specialty and role. In addition, the “thinking like an animal” has a place in the field of computer science and is used in programming or Soft Computing. In recent years, the field of swarm intelligence (SI) and adaptive behaviour (AB) has garnered a place in the field of robotics and artificial intelligence (AI) applications.

Today, Swarm robotics is one of the most popular research areas in robotic technology. It brings together experts in artificial intelligence, control theory, robotics, systems engineering and biology with the goal of understanding swarming behaviours in nature and applications of biologically-inspired models of swarm behaviors to large networked groups of autonomous vehicles. Many processes and subjects are interrelated to one and another namely; biology, artificial intelligence, robotics, organism behaviour, swarm architecture, vehicle model, modeling, analysis and synthesis. They are many ways to interpret and to prove the theoretical concept of swarming. These include mathematical analyses and virtually simulating the optimization scenario. SR systems also require proper controllers and sensing mechanism techniques. Thus, it is very important for researchers to design suitable platforms as test beds. This work also requires expertise from various fields such as biologists, system engineers, and experts on robotics, control system theory and artificial intelligence.

2 Swarm based AI applications

Most of animal-based metaheuristic algorithms are also related to SI. SI has gathered enormous research interest in related fields in recent years. SI is defined as “Any attempt to design algorithms or distributed problem-solving devices, inspired by the collective behavior of social insect colonies and other animal societies (Bonabeau et al. 1999).” It has also been mentioned by Karaboga et al. (2012), as a collective behaviour of centralized and self-organized system. Swarming behavior is one of the sub chapters under biological based or bio-mimicry or bio-inspired algorithm concepts. Swarm is a concept adopted from collective activities shown by a group of people or animal aggregating and forming an emergent behavior naturally. Three branches of the main applications which are frequently used for bio-inspired algorithm or SI are: (1) optimization and networks design, (2) prediction and forecasting, and (3) swarming robots (SR) (Fig. 1).

Fig. 1
figure 1

Among the application of bio-inspired algorithm

Full size image

To some extent, SR can also be classified under optimization design application. For instance, a group of flying drones moving in swarming formation in order to optimize the coverage area for each agent. However in this paper, it will be split up into another application since it requires not only the algorithm itself but also the physical setup and suitable design in order to execute and apply the bio-inspired algorithm movement strategy. This section focuses more on the Swarming Robots application as an overview for the new swarming robotics researchers. Thus, more examples will be highlighted in Sect. 2.3 compared with Sects. 2.1 and 2.2.

2.1 Optimization and networks design

The problem involves the process of finding a solution for an event under the conditions and certain restrictions. In practice, this means fulfilling the duties specified in the most efficient manner or producing the maximum output while resources are limited (Zain et al. 2011). Optimization problems are significant and occur extensively and exist in various forms in our daily lives. A bio-inspired form of networking enables further scalability and efficiency in communication and coordination in massively distributed systems architecture (Dressler and Akan 2010a). For example, suppose a school bus driver needs to provide transportation, to send students to their schools and to return home. Students’ homes are located in a housing area. They all need to be sent home on a route that can minimize the cost of fuel expenses and saves time. To do this, the bus driver should plan his trip; schedule a good sequence of which students who would want to be returned and sent first and which ones later. Factors such as narrow paths, avoiding traffic congestion, which is the shortest path direction, should also be considered. Another factor to be considered is to do without having to turn back on the same route more than once. In most cases, the best route can be found through experience.

Optimization and networks design applications are very broad, traversing from a specific area and ranging from engineering design (Schwabacher et al. 1998; Coelho and Mariani 2008; Kashan 2011), process optimization (Egea et al. 2010; Joshi and Pande 2011; Kwak and Kim 2012), scheduling system (Andersson et al. 2007; Frantzénl et al. 2011; Skobelev 2011), routing and flow control in networks and networking (Madan et al. 2007; Shakkottai and Srikant 2007; Minoux 2010), to service oriented applications in finance (He et al. 2008; Leibfritz and Maruhn 2009; Pennanen 2011), healthcare (Harrell and Lange 2001; Bagirov and Churilov 2003; José et al. 2011), and bioinformatics (Hernandez and Kambhampati 2004; Nebro et al. 2008; Arredondo et al. 2011). For example, in the formulation of the scheduling system, optimization can be used to determine the course of vehicle systems to the various destinations, to determine the scheduling of jobs in the factory, scheduling lectures at universities, creating timetable, computer network design and planning strategies for finding an optimal decision.

Vehicle routing problem (VRP) was introduced 50 years ago by Dantzig and Ramser. VRP studies have led to major developments in the field of heuristic and exact algorithms. In some very sophisticated algorithms, correctly parsing mathematical programming and metaheuristics for VRP authorities are developed in recent years (Laporte 2009). Originally, this problem is related to traveling salesman problem (TSP). TSP is NP-hard problem in combination optimization, studied in operations research and theoretical computer science. Given a list of cities and their pair-wise distances, the task is to find the shortest trip as possible in order to visit every city only once (Marinakis et al. 2011).

Figure 2 illustrates a different task but is still based on the same model, albeit in a different situation. Suppose there are van couriers such as in-service DHL Express. Each van’s itinerary is constantly changing every day and there are instances when the use of two vans would be better than three if the number of items out on that day is not so numerous. The process is called optimization. To produce satisfactory results, expertise is essential in analyzing, specifying and solving problems as well as the knowledge to select, make and use the correct approach.

Fig. 2
figure 2

The delivery system which requires a combination of methods of optimization and integer programming for an effective solution

Full size image

Although the human brain is supposedly to be more superior to animals in decision making, the animal’s way of thinking is appropriately suited for their life and environment. An ant which moves alone may seem lifeless and without a sense of destination. However, when they move in groups, the intelligence that these groups of insect demonstrate is bizarre. This sort of optimization approach seems more practical and simple. It is no wonder that lately, it has triggered interest among researchers.

2.2 Prediction and forecasting

The searching strategies are also designed to accommodate different situations, which are determined by historical information, (i.e. to predict or to forecast) (Lou et al. 2011). To mention a few examples; housing market fluctuations (Azadeh et al. 2012), trend adjustment for electricity demand forecasting (Wang et al. 2011), short-term food price forecasting in China (Haofei et al. 2007), financial forecasting (Kim 2004), short-time weather forecasting (Kilifarev et al. 2008), The model of rainfall forecasting (Zhao and Wang 2010), urban traffic forecasting model (Hong et al. 2007), forecasting output of integrated circuit industry (Pai et al. 2009), traffic safety forecasting (Gang and Zhuping 2011), simulating believable crowd and group behaviors (Huerre et al. 2010), Tawaf simulation for hajj training application (Rahim et al. 2011) and crowd modeling and traffic simulation (Lin and Manocha 2010).

If viewed in random, natural cycles that happen around people are actually closely aligned with each other. The movement of 10s of 1,000s of birds and fish swarm will produce clusters and shapes of certain formations. Based on the natural life cycle, the weather conditions which occur prior to the moment in time could also be reenacted through computer simulations databases.

Figure 3 shows the commercial software used for projecting future capital market expectations. By studying the movements of nature, it is no wonder that humans will also be able to anticipate other situations. The movement of a group can be used in applications, for example, the movement of pilgrim swarms (Fig. 4) during the hajj. The motion effect of the situation either on an individual basis or a group of pilgrims can firstly be seen through a simulation. Similarly, it is to determine the width of travel lanes and the appropriate size of the main gates in connection to the density of the pilgrims at a time.

Fig. 3
figure 3

ATS equity engines by adaptive trading systems

Full size image
Fig. 4
figure 4

Simulation of congestion during the Hajj. Photo taken from Narain et al. (2009)

Full size image

2.3 Swarming robots

The concept of a swarming robot system is adopted from nature; from the appearance of flocking birds, the movement of a school of fish, the ant colonies and swarming bees among others. This “emergent behavior” is the aggregation result of many simple interactions occurring within the animals themselves (Corner and Lamont 2004). It is a broad field of computational swarm intelligence and is applied in swarming robotics (Jevti and Andina 2007). A group of ants is able to defeat a centipede which is 10 times larger than an ant. A group of successful flies can easily locate their favorite food. The formation movement of a flock of birds has been copied by the air force in the battlefield and often appears in air show during the National Day Celebration. Figure 5 shows the aircraft formations that resemble the movement of a flock of birds. Unlike biomechanical robots, the design is not necessarily the same as the original species itself in swarming robots. The most important factor for design is the formation of the movement, the searching strategy, how consensus is achieved and the decision made.

Fig. 5
figure 5

a Formation movement of aircraft. b Formation of bird movements. Photo courtesy of Phill Price via the Londonist Flickr Pool

Full size image

Nature has demonstrated that, like humans, all living creatures including animals also have their needs for survival in life. Most animals do not need to move away or change their habitats if sustenance is available. The phenomenon of animals moving in groups shows the existence of a social community environment. Even as humans, we need to be in a circle of social network in order to attain a more well-informed community.

Although modeling bio-inspired robots and biology have made important contributions to both research and robotic insects, insects and robots are still separated by a gap between the living and the merely mechanical (Sharkey 2006). Practically, autonomous robots require the behavior of robots solely on their local perception, which is usually quite limited. If the swarm of robots participate in a variety of examples based on individual perceptions of a global picture, then the swarm can perform efficiently and can target a group of complex tasks (Schmickl et al. 2007). State of the art technology such as swarm robotics is an algorithm broadly applicable in a wide area due to factors such as self-organization, robustness and flexibility. However, high cost at the initial stage and during operation are the main obstacles in the multi-robot foraging operating system (Lee and Ahn 2011). The core idea behind this theory is the ‘Nature’s Behavior’ such as, swarm of bees, ant colonies, bird flocking, fish schooling, bacterial growth, termites and many more. Typically, these systems consist of a simple agent population with a very small brain and duties of local interaction. However, collectively, they make a very systematic approach in their search missions globally. In general, the system consists of agents who adhere to very simple rules, such as finding food. There is no command or centralized control to tell them where to go and how to do it. Individual agents should behave according to a certain local degree and information is passed randomly among them. This action led to the emergence of organized systems and teamwork, but it is not known by the individual agents.

The basic criteria in designing the swarm robot’s hardware system are; mechanical design, dynamic and kinematic design, communication, sensory system and power management. The design of the components is dependent upon the mission and the biological behaviour itself. Artificial intelligence is mainly related to the software portion, where the robots are programmed to imitate the ‘Nature’s Behaviour’. The costs of the robots are also critical. Most of the swarm projects consist of 4 units robots, and it can grow up to 100 units and above depending on the application itself. For some research groups, the ideology has already materialized. Some projects as shown in subsequent figures.

Figure 6a shows the Formica Robots which are controlled by an MSP430 microcontroller. Each agent has two mobile phone vibration motors directly driving the wheels of small neoprene. The goal is to build open source swarm robots which are as cheap as possible (English et al. 2008). Figure 6b is the heterogeneous swarm robotics for search applications (ARGoS) which was developed within the Swarmanoid project. It is a DARPA project in which the goal is to show a large number (100+) of heterogeneous physical robots working together to solve indoor search applications. This project is a joint venture between Science Applications International Corporation (SAIC), University of Tennessee, Telcordia Technologies, and University of Southern California (Caro et al. 2010).

Fig. 6
figure 6

a The formica robots. b Swarmanoid project

Full size image

The REPLICATOR project (Fig. 7a) focuses on the development of advanced robotic systems, which consist of a large-scale super-small micro-autonomous robotic mobile that can install itself into large artificial organisms. This robot organisms have a common energy and reliable information as well as bus-legged, wheeled or climbing locomotion, based on a modular sub-systems which can be autonomously reconfigured (Levi and Kernbach 2010). Figure 7b shows the I-Swarm project (Jasmine), a cooperation project between the University of Stuttgart (Electronics Division) and the University of Karlsruhe (Mechanical). The project is to produce robots as small as 3 cubic cm only in size. These robots are able to unite and form unusual formations.

Fig. 7
figure 7

a REPLICATOR project. b I-Swarm project (Jasmine)

Full size image

UAV SWARM Health Management Project (Fig. 8a) is supported by Boeing Phantom Works. To investigate the techniques that will enable the execution of continuous (24–7) mission operations using multiple autonomous vehicles (i.e. vehicle SWARMS) in a dynamic environment (Valenti 2007). Project distributed flight array (Fig. 8b) is made up of individual moving robots on the ground. It is actually a flying platform which consists of a number of various autonomous vehicles, capable of driving a single propeller, dock with their ‘friends’, and fly regularly. When the hovering flight have covered a certain range for a few minutes, they then fall back to the ground, only to repeat the cycle again (Oung et al. 2010).

Fig. 8
figure 8

a UAV SWARM health management project. b Project distributed flight order

Full size image

The SMAVNET project (Fig. 9a) aims at developing swarms of flying robots that can be deployed in disaster areas to rapidly create communication networks for rescuers. Flying robots are interesting for such applications because they are fast, can easily overcome difficult terrains, and benefit from line-of-sight communication (Hauert et al. 2010). Another example is the robot car project of Nissan EPORO by a group of robots mimicking the fish movement (Fig. 9b). This project is intended to avoid accidents and traffic congestion for automatic car driving in the future.

Fig. 9
figure 9

a SMVNET project. b Nissan EPORO

Full size image

Another interesting application is the Swarmbots (Fig. 10a). In this system, a group of robots work together so as to build a self-bridge to get across, then down into the rescue hole and finally managed to pull out the object equivalent to a child’s weight (Ampatzis et al. 2009). Kilobot (Fig. 10b), is a low-cost robot designed to make the testing of collective algorithms on hundreds or thousands of robots accessible to robotics researchers. To allow for the possibility of a collectively large number of robots whose order of magnitude is greater than the number that exists today, each robot is made with low-cost parts and takes 5 min to assemble. This project is funded by WYSS Institute for Biologically Inspired Engineering and NSF (Rubenstein et al. 2012).

Fig. 10
figure 10

a Swarmbots. b Kilobot

Full size image

3 Basis of framework for swarming robotic

Mimicking the behavior and perception of animals is the key element in the development of swarming robots. Although the design of the robots appears too simple and utilizes low cost computing techniques, it is, however, necessary for the design to have a few robots that are intelligent enough to tackle the optimization problem. It may sound ironic that a swarming system does not seem to require high speed processing, big memory space and complex programming to execute its mission; nevertheless, the collective work outcome of the robots as a team is surprisingly effective such as in the case of the ants in the real world.

Vijay Kumar had introduced the framework on how to develop an engineering system inspired by the swarming behavior of animals (Kumar 2010). Figure 11 shows that the proposed methods can be categorized into 3 phases:

  1. (1)

    Phase 1 (biological/behavior) to understand the biological behavior of the particular animal,

  2. (2)

    Phase 2 (algorithm/simulation) to analyze the behavior in the form of algorithm and computer simulation of the motion.

  3. (3)

    Phase 3 (engineering/applications) to alter the theory for a specific engineering application.

Fig. 11
figure 11

A framework and methodology for the analysis of swarming behavior in biology and the synthesis of bio-inspired swarming behavior for engineered systems

Full size image

This methodology is expected to provide insight on how to understand the term an “animal inspired engineering system”. As mentioned earlier, it is a multi-disciplinary project where it requires expertise from various fields. The main mission is to study the natural behavior and the biological application in the optimization problem in an automated robotic system network. The framework and methodological approach to understand the engineering system designs that are necessary are as follows:

  1. (1)

    modeling the motion behavior naturally,

  2. (2)

    analyzing the swarming behavior and emergent network formed and

  3. (3)

    synthesizing the inspiration of biological formations.

The new theory of optimization algorithm can also be modeled through the robotics platform. SR is also related to metaheuristic algorithm and artificial intelligence (AI). Each individual robot is also known as an agent. It follows a very simple rule and easy task. There is no leader, nor there is any specific order given. Swarming results can be seen naturally as a result of the whole process. At this particular stage, each agent still goes on performing its basic task without knowing that the main group’s objective has already been achieved.

In general, the term biological based artificial intelligence refers to the concept of a group of bio-mimetic and is sometimes associated with bio-mechanics. As presented, this approach is inspired by biological properties, mostly in connection with the adoption of a community network based on the social intelligence of animals. It also part of the field of programming and intelligence branches in relation to meta-heuristic algorithm. Each individual in the group, known as agents act only in accordance with a simple instruction. No one acts as the head of the ruling class or gives specific instructions. Group formations exist in a spontaneous manner. In that event, each agent itself still continues to perform their individual tasks involuntarily and trapped in the close loop system.

Among the factors identified why animals moving in the group is due to the following motivation: find a better source of food, nomadic, mating, the movement to save energy and space, survival, self-defense, attacking the enemy, nest or place of residence and co-operation. The last factor yielded better results. Generally, the basic concept and principles for each agent is identical, which is as follows:

  • Each agent must have the same degree of capability, strength and speed. This is in accordance to the nature of animals and insects. If not, the swarming formation will be interrupted. In case one agent fails, it will not affect the overall movement and mission.

  • The whole system is automated and there is no leader. In an actual robotic system, a server may be required but this is only to serve as a communication medium. Each agent acts upon the instruction given by its own brain (micro-controller) and natural behavior. They carry out their task on their own accord. The intelligence can only be seen in the group’s formation.

  • Communication medium varies according to the biological species itself; e.g. the ant by its chemical pheromone, the bees by its wangle dance and the bat through its sonar capability.

  • Each system has its own organizational family.

  • Each agent performs the same objective, knows its natural task; either in attacking mode, collecting food or building a nest.

  • Each act occurs in a short period of time. No energy is wasted. Each agent consumes only the sufficient amount that it requires, nothing in excess. This is the natural optimum approach.

  • The individual insect does not have a smart and large memory. No mathematically complex calculation is done. They merely follow the estimated direction and lesser accuracy does not impede their performance.

In short, the agent in the specie can never be trained like the cat and the dog where the location of food is known prior to that or is based on a previous exercise or memory. In swarming robotics, parameters such as memory and processing time are very limited. Having seen the efforts carried out by recent researchers in other places, it is time to see the approach taken to design a system based on robotic swarm of fruit flies that have been implemented (Abidin et al. 2012).

4 Conclusions

This chapter seeks to provide a basic understanding on the roots of the concept of natural pattern formation or better known as SI for SR application. SR and SI are one of the important branches in Artificial Intelligence subject. The state-of-the-art techniques in connection to this subject were reviewed.