Proceedings of International Joint Conference on Neural Networks, San Jose, California, USA, July 31 August 5, 2011

A Distributed, Bio-Inspired Coordination Strategy for Multiple Agent Systems Applied to Surveillance Tasks in Unknown Environments
Rodrigo Calvo, Janderson Rodrigo de Oliveira, Mauricio Figueiredo and Roseli Ap. Francelin Romero

Abstract Multiple agent systems are applied to exploration and surveillance tasks. A new distributed coordination strategy, designed according to a modied version of the articial ant system, is described. The strategy is able to adapt the current system dynamics if the number of robots or the environment structure or both change. Experiment simulations are executed to evaluate two versions of the strategy considering different multiple robot systems and environment structures. Results conrm that exploration and surveillance general behaviors emerge from the individual agent behavior. Different compiled data sets are considered to assess the strategies, namely: needed time to conclude the task; and time between two consecutive sensory on a specic region. The results show that the strategy is effective and relatively efcient to execute the exploration and surveillance tasks.

I. I NTRODUCTION Multiple agent systems are well characterized if the system dynamics reect some synergy, that is, from the individual behaviors emerges a global one that becomes the system capable to reach a specic goal. If only one agent of a group achieves equally the same goal with the same performance the entire group does, then the agents, at rst, are not a multiple agent system. There are many applications to which multiple agent systems are the suitable approach to be adopted, such as: rescue operations in catastrophic events; re extinction; and exploration in hostile environment [1], [2], [3]. Some of the main reasons that justify this choice, among others, are: great dimension of the task and reduced resources (e.g.: velocity, strength, energy) provided for a single agent; necessity to adaptation to spatial or temporal variation of service demand; and robustness. For some tasks this approach is mandatory; for others it is a matter of convenience to increase the quality, to improve the performance or to save monetary funds. Nowadays the technology reaches more sophisticated levels providing environment support and embedded supplies. Those improvements bring closer the possibility of multiple agent systems to become usual. The strong expectation associated to this possibility captivates the attention of the scientic community. Different aspects are investigated in
Rodrigo Calvo, Janderdeson Rodrigo de Oliveira and Roseli Ap. Francelin Romero are with the Department of Computer Sciences, University of Sao Paulo, Avenida Trabalhador Sao-Carlense, 400, P.O. Box 668, 13560-970, Sao Carlos - SP - Brazil (phone: +55 16 33739661; email: {rcalvo,jrodrigo,rafrance}@icmc.usp.br ). Mauricio Figueiredo is with the Department of Computer Sciences, Federal University of Sao Carlos, Rodovia Washington Luis, 235, P.O. Box 676, 13565-905, Sao Carlos - SP - Brazil (phone: +55 16 33518232; email: redo@dc.ufscar.br ).

multiple agent systems, such as: agent communication and information merging [4], [5], [6], [7]. Another important aspect is the agent coordination that allows the system accomplishes efciently general tasks such as: exploration, coverage, surveillance, among others. On the one hand, coordination strategies are designed to provide multiple agent systems with a set of characteristics, e.g.: decentralized coordination, small redundancy of agent efforts, and strong cooperative behavior. On the other hand, designers devote effort to concept coordination strategies that are dependent on the least number of parameters as possible. A tricky parameter is the number of agents. Another requirement that may depreciate the strategy is the total knowledge of the environment. According to a technique described in [8] robots construct a common map cooperatively. The author introduces the notion of a frontier, which is a boundary between the explored and unexplored areas. Along robots move, new boundaries are detected and frontiers are grouped in regions. Then robots navigate toward the centroid of the closest region, while sharing maps. The strategy is a centralized type since the A algorithm considers all information that the robots provide and the algorithm output denes the next steering direction of each robot. The strategy does not avoid unnecessary redundancy of robot efforts. Stigmergy elds based methods for cooperation have been recently employed in the context of robotic exploration [9], [10], [11]. They rely on a mechanism of indirect communication among the agents which allows their actions to be inuenced by a trace left previously in the environment by the robot. In this way a task can be accomplished in an efcient manner. A different coordination scheme in [12] is presented based on potential elds in which repulsive forces repel robots from each other and obstacles. Starting navigation from the same region, the robots keep moving until repulsive forces cancel each other. At this moment the sensor network is settled and the robots stop. This approach ensures the coverage of the whole environment if the number of robots is sufciently great. Coverage tasks are the focus of the investigation in [13]. The distributed coordination strategy, based on the Voronoi diagram and Delaunay triangulation, is proposed to maximize the coverage area. The strategy is robust to robot failures. Voronoi diagram is also adopted to solve a coverage problem in [14]. Despite these strategies solve a coverage problem, both do not cover completely the environment.

978-1-4244-9636-5/11/$26.00 2011 IEEE

3248

The task of coordination of multiple agents is considered complex [15], [5]. Coordination strategies based solely on mathematical formulation; and agent and environment models are very parameter dependent and suffer critical degradation due to agent failure [16], [13], [14]. Bio-inspired and evolutionary theories provide fundamentals to design alternative strategies [17], [18]. Particularly, the articial analog versions of biological mechanisms that dene the social organization dynamics, observed in some swarm systems, are very appropriate in applications involving multiple agents, for example, decentralized control, communication and coordination [19], [20], [21]. A new coordination strategy, named Inverse Ant SystemBased Surveillance System (IAS-SS), designed according to a modied version of the ant algorithm is described [22]. The strategy is primarily for coordination of multiple agents applied to surveillance and exploration tasks. Some characteristics of the strategy IAS-SS are: decentralization, parameter independence of the number of agents, and also independence of the environment structure. The strategy is able to adapt the current system dynamics if the number of robots or the environment structure or both change. Experiment simulations are executed to evaluate two versions of the strategy IAS-SS considering different multiple robot systems and environment structures. Results conrm that exploration and surveillance general behaviors emerge from the individual agent behavior (move to where there is less pheromone). Another emergent behavior observed is the obstacle avoidance (there is no specic mechanism that generates this behavior). A totally uniform strategy is considered to comparison purposes. This strategy is able to execute neither the exploration nor the surveillance tasks. Different compiled data sets are considered to assess the strategies, namely: time to conclude the exploration task; and time interval between two consecutive sensing of any specic region. The results show that the strategy is effective and relatively efcient to execute the tasks. The remainder of the paper is organized such as follows. Section II provides fundamentals of the articial ant system theory. The description of the multiple robot system for exploration and surveillance tasks and the coordination strategy IAS-SS are the focuses of the Section III. The dynamics corresponding to pheromone releasing, pheromone evaporation, and the mechanism to determination of steering direction are also dened. Section IV shows simulation results obtained from a set of experiments. The main contributions and relevant aspects of the paper as well as expectations for future works are highlighted in Section V. II. A NT S YSTEM Surprisingly the complex tasks that ant colonies perform, such as object transportation and build edges, demand relatively more capabilities that a single ant is endowed [23], [24]. Biological ants have two known mechanisms to establish communication, namely, direct and indirect. Biological

ants not only exchange stimuli when they meet; but also exchange stimuli indirectly (a communication mechanism called stigmergy). Ants deposit a specic type of substance (pheromone) on the ground while they move. There are different types of pheromone, each of which associated with a particular meaning. If a pheromone trail is found and this pheromone type indicates food, then more and more ants follow this trail, depositing more pheromone and reinforcing the stimuli. An opposite behavior happens if the pheromone is of the aversive type, indicating risk and danger. Stigmergy mechanism is considered one of the factors that decisively contribute to amplify the capabilities of a single ant. Ant colonies use the stigmergy mechanism to coordinate their activities in a distributed way [25]. Articial ant systems are the articial counterparts of the biological ant colonies, designed to solve complex problems, among others: optimization combinatorial problems [22]. Analogously articial ants (e.g.: robots) are able to use the stigmergic communication. Pheromone trail provides a type of distributed information that articial agents may use to take decisions or modify to express previous experiences [26]. A distributed coordination behavior emerges from this capability, providing solutions to problems associated with exploration in hyperspace. III. I NVERSE A NT S YSTEM -BASED S URVEILLANCE S YSTEM (IAS-SS) The system proposed is for exploration and surveillance tasks and it is designed according to the main ideas of the articial ant system. Essentially the system is a group of articial agents (e.g.: robots) each of which moving independently and taking decisions based on the stimuli received only from the environment. While the robots navigate they deposit a specic substance, the pheromone (the analogue of the pheromone in biological ant systems), into the environment. At each time each robot receives stimuli from the pheromone and adjusts its navigation direction. This is the only one decision that a robot takes. In fact, the robot navigation system considers a set of stimuli detected at different angles and same distance. The lesser is the detected amount of the substance the greater is the probability that the robot takes the navigation direction equal to the angle where this amount of substance is. The logic of the decision in the IAS-SS is the opposite of that adopted in the traditional ant system theory. The logic adopted there generates a positive feedback, that is, the greater the amount the substance the greater is the probability of the agent to follow the respective direction. The block diagram in Figure 1 represents the sequence of main actions that an agent system performs at each iteration.

Fig. 1.

Functional Diagram Block for a single agent.

3249

It is important to mention that the robots exhibit the obstacle avoidance behavior, but there is no specic embedded navigation mechanism for that. In fact, this navigation skill emerges from the synergy among the articial agents as a natural consequence of how the pheromone is released on the environment and the effects the pheromone stimuli generate. A detailed description of the IAS-SS system is given in the next. Consider a group of N robots k, k = 1, . . . , N . Every robot k performs two basic operations: steering direction adjustment and pheromone deposition. A. Steering Direction Adjustment Two strategies to determine the steering direction angle are adopted. The rst, Stochastic Sampling, considers all pheromone stimuli that the sensor detects at the border of its range (Figure 2). The second, Best Ranked Stochastic Sampling, determines the adjusting of steering angle based on only those stimuli associated with the least amount of pheromone. The model of the sensor adopted is such that it detects pheromone stimuli at a specic distance R, from 90 degrees to the left to 90 degrees to the right of the steering direction, corresponding to the average of the amount of pheromone deposited in an angle interval. The total range of 180 degrees is divided in identical angle intervals, such that the sensor detects stimuli corresponding to different angles As , s = 1, ..., S.

According to this strategy robots tend to move to directions where there is low amount of pheromone. The general behavior observed is that the robots move to unexplored areas or areas barely visited by robots during some period of time. The adjusting of steering direction is given by: k (t) = k (t 1) + A(s ) (2)

where k (t) is the steering of movement of robot k at instant t, [0, 1] is the constant coefcient for smoothing of steering direction adjusting and A(s ) is the selected direction by probability of equation 1. However, Stochastic Sampling mechanism is not efcient for large areas where the amount of pheromone deposited is similar on every point. Even if the amount of pheromone differs a bit, it is possible to choose angles As that dene bad steering directions due to the stochastic nature of the strategy. To maximize the explored area in a reduced period of time, it is investigated a second mechanism, described below, that is a different version of the Stochastic Sampling. 2) Best Ranked Stochastic Sampling: Differently from the Stochastic Sampling, not all angle intervals are considered to dene the steering direction, but only two subsets of them. The rst, the angle intervals are those that the amount of pheromone is very low. Specically, the strategy sorts the intervals according to the respective amount of the pheromone. Then only those angles As associated with the least amount of pheromone are considered to dene the steering direction. The second subset consists of elements chosen randomly, according to a uniform distribution, from the angles As that are not in the rst subset. B. Pheromone Releasing and Evaporation In traditional articial ant systems, agents release pheromone on the ground only on their respective positions signaling exactly the robot way [22]. Differently, the articial agents in the IAS-SS spread out pheromone on a wide area in front of their respective positions, corresponding to sensor range area. Once the agent determines the steering direction (see Section III-A), but before it moves to, it spreads pheromone. The amount of pheromone deposited on the ground decreases as the distance from the robot increases. The model for the pheromone releasing is such as follows. Consider that Lt be the sensor range area at iteration t and Q the entire environment space, such that, Lt Q R2 . Then the amount of pheromone k (t) deposited by kth robot deposited at the X position X at iteration t is givem by: k = X
2 , if X Lt e 0, otherwise (XXk )2

Fig. 2.

Robot and sensor models

1) Stochastic Sampling: A probability value is assigned to each discrete angle in the sensor range. The probability assigned to the angle As is inversely proportional to the amount of pheromone deposited in the respective angle interval, that is, the lower is the amount of pheromone detected, the higher is the probability associated with the angle. Specically, the probability P (s) assigned to the angle As is: P (s) = 1 s /
S i=1 i

(3)

(1)

where s is the amount of pheromone corresponding to the angle As . The adjusting of steering direction is determined according to a discrete random variable a dened through the probability P (s), assuming values in the set As , s = 1, ..., S.

where Xk is the position of the kth robot and is the Gaussian dispersion function. Multiple robots deposit pheromone in the environment at same time, then the total amount of pheromone deposited at the position X at iteration t depends on the contribution of every robot.

3250

Furthermore, pheromone is not a stable substance, that is, it evaporates according to a specic rate. The total amount of the pheromone that evaporates X (t) at position X and time t is modeled such as follow: X (t) = (1 )X (t) (4)

where is the evaporation rate and X (t) is the total amount of pheromone on the position X at iteration t. Therefore, the total amount of pheromone X (t) at X and at time t is (Equation 5):
K

X (t) = (t 1) +
k=1

k x,y

(5)

IV. E XPERIMENTAL R ESULTS Experiment simulations are developed to evaluate preliminarily the bio-inspired coordination strategy of IAS-SS. The strategy is considered to generate the dynamics of multiple robot systems applied to exploration and surveillance tasks. The Player/Stage platform 1 is used to perform the experiments. The Player/Stage is a robot server designed by the University of Southern California 2 for distributed control. Player operates in a client/server environment and the communication between them occurs through TCP/IP protocol. Stage is a simulator for robots and sensors for two-dimensional environments. Player/Stage models various robots and sensors simulating simultaneously their exact dynamics, including odometer error models. For the purpose of the experiments, the robot Pioneer 2DX is chosen to be modeled in the Player/Stage platform. This robot is equipped with a laser range-nder able to scan the environment (general obstacles, e.g.: walls and objects). The experiments are arranged in three groups. The rst consists of experiments focusing on the steering direction mechanisms described in Section III-A. The mechanisms are compared with a completely uniform one. The second group of experiments is designed to investigate the inuence of the conguration of robot initial positions in the task performances. The experiments in the third group aim at analyzing the impact of the number of robots on the system performance. The experimental data are selected and compiled assuming the following meaning. First, the exploration task is executed if the environment is completely covered, that is, the system is capable to provide information to map the environment completely. Moreover, the faster the system completes the task, the better is the performance. Second, the system carries out the surveillance task if there is no instant T* such that after this instant exists a region that is not sensed anymore. Despite this denition for surveillance task is accurate, it is not suitable since may be impossible to nd T*. Therefore, for practical purposes, it is important that the system conclude the task continually, that is, the system has to be able to
1 http://robotics.usc.edu/player 2 www.usc.edu

sense the entire environment considering that a new sensing task is started when the system concludes the previous one. Furthermore, the lesser is the maximum time between two consecutive sensing tasks, the better is the performance. The environments where the IAS-SS system carries out tasks are divided in connected small regions called here rooms. The system parameters used in the experiments are: Pheromone releasing and evaporation dynamics: = 0.43R (radius of the semicircle where the pheromone is deposited, see Figure 2); = 0.01 (evaporation rate); and X (0) = 0.5 (the amount of pheromone at iteration t = 0). Robots and sensors: R = 8.00 meters (radius of the semicircle where the pheromone is deposited, see Figure 2); = 0.5 (constant coefcient for smoothing of steering direction adjusting); and Robot speed: 0.5 meter per second. Steering direction mechanisms: S = 360 (number of angle intervals). Simulation parameter: Maximum number of iterations = 1000. These parameter values correspond to those that the multiple robot system reaches the best performance, considering all previous experiments executed. The discretd time is adopted in simulation and it is equivalent to the number of iterations. A. Uniform versus Stochastic versus Best Ranked Stochastically Sampling Both steering direction strategies, Stochastic Sampling (SS) and Best Ranked Stochastic Sampling (BRSS), have profound random characteristics, since the steering direction adjustment is determined according to a discret random variable. In order to show that the respective performances are not a mere consequence of a random behavior, the strategies are compared with a uniform strategy (US). According to US strategy, a discret random variable, dened by an uniform distribution in the space of the angles As , determines the steering direction adjustments. Observe that there is no connection between the pheromone and the uniform strategy; different from SS and BRSS strategies. The environment designed for evaluation is such as in Figure 3. It is possible to identify six rooms. Three robots k, k {1, 2, 3}, start the navigation in the rooms 1, 6 and 2, respectively. Three sets of graphics, showed in Figure 4, summarizes the simulation results, each of which corresponding to a different strategy. For each strategy, three graphics are presented, each of which registering the behavior of one of the robots. The y-axis represents the rooms and the x-axis represents the iterations. Each vertical line indicates the iteration when the IAS-SS senses the entire environment (the robots visit cooperatively all the 6 rooms), considering that a new sense

3251

Fig. 3.

Environment structure

task is started after the system conclude the earlier one. The period between two vertical line, that is, the period when all rooms is sensed is denoted as cycle of Surveillance Interval (or S.I. cycle). Two aspects are considered for analysis: the time necessary to conclude the exploration task; and maximum time interval between two consecutive sensing of any specic region (S.I. cycle). Firstly, let us consider the exploration task. The graphics show that IAS-SS system with Uniform Strategy is able to conclude the exploration task, but after a long time, precisely at the iteration 959. Observe that with SS and BRSS strategies, IAS-SS system executes more efciently the task, that is, the system concludes the task very earlier, at the iterations 62 and 41, respectively. IAS-SS system with US strategy concludes the surveillance task only once (there is only one S.I. cycle), considering all the simulation. There is a strong contrast if this performance is compared with those obtained with the SS and BRSS strategies. Vertical lines indicate that the system with these strategies continually concludes the surveillance tasks (all the robots cooperatively visit the 6 rooms). The tables I and II summarize the data. The system with SS strategy concludes the surveillance task 9 times and BRSS 12, and the maximum intervals between two consecutive conclusions are 208 and 119 iterations, respectively. IAS-SS system with BRSS is clearly superior. The SS and BRSS strategies induce a stronger collaborative robot behavior than in the case of US strategy. Observe that robots in the pheromone dependent strategies vary more the rooms that they visit than robots do in the case of US strategy. For example, consider the robot 2. It visits the rooms 6, 4, 5 if IAS-SS is with US; 1, 4, 5, 6 if SS strategy is adopted; and the robot 2 visits all rooms in the case of BRSS strategy. B. Initial position of robots This group of experiments evaluates the efciency of IAS-SS for distinct conguration of robots regards as their positions. Two cases of conguration of robots are designed in order to analyze the system performance: 1) together conguration robots start navigation at same region (or room) and 2) separated conguration at distinct rooms. For exploration and surveillance tasks, it is obvious the greater efciency is guaranteed when the robots are not closer. However, experiments intend to demonstrate that, after a

(a)

(b)

(c) Fig. 4. IAS-SS performance according to different strategies: (a) US; (b) SS; (c) BRSS mechanism.

3252

TABLE I P ERIOD OF S URVEILLANCE I NTERVALS FOR S TEERING D IRECTION M ECHANISMS

Mechanism US SS BRSS

Max. Surveillance Interval 959 208 119

1st 959 62 41

2nd 151 119

3rd 116 98

Cycles 4th 48 64

of Surveillance Interval (iterations) 5th 6th 7th 8th 9th 10th 94 51 208 172 75 51 78 89 101 87 100

11th 61

12th 55

TABLE II M ONITORED ROOMS AT EACH C YCLE OF S URVEILLANCE I NTERVAL

Mechanism

Robot # # # # # # # # # 1 2 3 1 2 3 1 2 3 1st 1,4 6,4,5 2,3,4 1,4 6,4,5 2,3,4 1,4,5 6,4 2,3,4 2nd 5,4,1 5 4,6,3,2 5,4,1 4,3,2 4,6 3rd 4,1 5,4,6 2,3 1,4 4,5 6,4,3,2 4th 1,4,5 6 2,3 4,6 5,4,1 2,3 5th 5,4,1 6,4 2,3 6,4,5 1,4 3,4,2

US

SS

BRSS

Monitored Rooms 6th 7th 1,4,5 5,4,3,6 6,4 4,1,5 2,3,4 4,3,2 5,4,6 6,4,5 4,1 4,1 2,3,4 4,3,2

8th 6,4,1 4,6,5 2,3,4 5,4,6 4,5,1 2,3,4

9th 1,4,6 5 3,2 6,4,1 4,5 2,3,4

10th 1,4 5,4,3,2 4,6

11th 1,4,5 2,3 6,4,3

12th 5,4,1 3,4,2 3,4,6

while, scenarios of conguration of together conguration can achieve the same efciency of separated conguration. Experiments to analyze the performance according to initial position of robots is accomplished in environment of Figure 5. For both of cases of conguration, three robots are launched. In particular for separated and together congurations, they start navigation at rooms 2, 6 and 7; and room 6, respectively.

Fig. 5.

Environment structure

For next experiments, six environment congurations are generated from combination of cases of congurations (separated and together) and steering direction mechanisms of experiments of Section IV-A. The key of surveillance task is minimizing the time (iterations) which a region is not monitored. Hence, here, a manner to measure the system performance is analyze the maximum period (maximum number of iterations) which each room is unvisited. Maximum period of rooms is presented in Figure 6 for the six environment congurations. Although separated conguration presents slightly advan-

Fig. 6.

Maximum number of unvisited rooms iterations

tage over together conguration because, since iteration t = 0, three robots monitor three different rooms, the performance of both of congurations are similar. One of main characteristics of IAS-SS system is the skill of robots to keep distance from each other according to aversive pheromone. Then, even with together conguration, as long the robots move, they are spread in environment. Thus, the performance of together conguration becomes similar to

3253

separated conguration. That is, the advantage of separated conguration is diluted during navigation. To illustrate it, graphics of Figure 7 show the behavior of robots and surveillance intervals for separated and together congurations using BRSS mechanism. Other mechanisms are not presented due to limited space. C. Number of robots This problem discusses about the relation between the size of environment and number of robots. Indeed, higher number of robots is, more regions are explored and monitored simultaneously so that, few or no regions are empty for long period. Since robots behavior is based on inverse of ant algorithm, the probability of one robot explorer and monitor large environments is higher. However, it may take a long time. In order to evaluate the performance of motion coordination and the efciency of surveillance task, experiments are carried out with an increasing number of robots in environment of Figure 8. Because BRSS mechanism presented better performance than US and RS mechanisms in previous experiments, it is adopted to analyze the efciency of exploration and surveillance tasks while the number the robots increases. All robots added are placed at room 1. Although it is clear that the time to explore decreases as number of robots increases, the surveillance task is accomplished even with a number restricted. It emphasizes that number of robots is not a factor to limit the size of the explored environment. Even few robots are able to monitor large areas. The S.I. cycle is completed independently of number of robots. However, as general behavior of system, the length of S.I. cycle is reduced while the number of robots increases. Also, as consequence of addition of robots, there are more completed S.I. cycles. It can be observed in Table III.
TABLE III P ERFORMANCE OF TOGETHER CONFIGURATION Number of robots 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Number of S.I. 3 5 6 7 9 12 13 14 19 21 21 22 26 25
AND

(a)

BRSS

Average of S.I.(iterations) 211.5 178.75 147.2 117.3 93.625 73.6364 58.0833 57.5714 44.5 38.9 39.15 39.1429 31.96 33.2083 (b)

V. C ONCLUSIONS AND F UTURE W ORKS This work describes a new bio-inspired distributed coordination strategy, named IAS-SS, for multiple agent systems applied to exploration and surveillance tasks. The strategy is based on a swarm theory, specically the ant system theory. According to this strategy, the agents are able to indirect

Fig. 7. IAS-SS performance according to different congurations for BRSS: (a) separated; (b) together congurations.

3254

Fig. 8.

Environment structure

communication as the biological agents are, but their reaction to the pheromone is distinct. The IAS-SS strategy denes steering directions that guide preferably the agents to where the amount of pheromone is lesser. A set of experiments are proposed for performance analysis. Experiments consider two performance criteria: time for conclusion of exploration task and maximum time between complete environment sensing (surveillance task). Two parameters, namely: start position, and number of robots, stress the strategy capabilities. Two versions of IAS-SS strategy are considered and compared with a totally random strategy. The IAS-SS strategies are signicantly superior. Some characteristics of these strategies are preliminary conrmed: they are not dependent on the knowledge of the environment structure; they are robust in regard to the number of robots, that is, in case of the number of robots changes (due to failure or insertion), these strategies are able to change the system dynamics in order to reach a good performance; and these strategies keep robots well separated guiding them toward regions not recently visited. As future works some parameters of IAS-SS system will be considered for analysis, e.g.: the pheromone releasing mechanism. Moreover, a localization method will be integrated to IAS-SS system in order to deploy it in real robots. In this case, a chemical sensor will be attached to the front of robot. Similarly, a device to disperse the chemical will be deployed. A more simple way is to consider only distance sensor and set the cells of built map to indicate that there is amount of pheromone at respective position. In addition, more complex surveillance tasks, e.g., those that a strange agent invades the environment, will be investigated. ACKNOWLEDGMENT The authors would like to thank FAPESP and CNPq for their support. R EFERENCES
[1] J. G. Bellingham and M. Godin, Robotics in Remote and Hostile Environments. Science, vol. 318, pp. 1098-1102, 2007. [2] H. H. Schmitt, From the Moon to Mars. Nature, vol. 301, pp. 36-43, 2009. [3] F. Mazzini, D. Kettler, J. Guerrero and S. Dubowsky. Tactile Robotic Mapping of Unknown Surfaces, With Application to Oil Wells. IEEE Transactions on Instrumentation and Measurement, vol. 60, pp. 420429, 2011. [4] X. L. Long, J. P. Jiang and K. Xiang. Towards Multirobot Communication. IEEE International Conference on Robotics and Biomimetics, 2004. ROBIO 2004, pp. 307-312, 2004.

[5] A. Speranzon. Coordination, Consensus and Comunnication in MultiRobot Control Systems. PhD Thesis, Royal Institute of Technology, Stockholm/Sweden,2006. [6] C. Liu, Y. Ma and C. Liu. Cooperative Multi-robot Map-Building Under Unknown Environment. Proceedings of the 2009 International Conference on Articial Intelligence and Computational Intelligence, vol. 3, pp. 392-396, 2009. [7] L. Andersson and J. Nyg rds. On Multi-robot Map Fusion by Intera robot Observations. In proceedings of 12th International Conference on Information Fusion, 2009. [8] B. Yamauchi. A frontier-based approach for autonomous exploration. Proceedings of the 1997 IEEE International Symposium on Computational Intelligence in Robotics and Automation, pp. 146-151, 1997. [9] D. Scheidt and J. Stipes. Cooperating unmanned vehicles. Proceedings of the 2005 IEEE Networking, Sensing and Control, pp. 326-331, 2005. [10] J. Stipes, R. Hawthorne, D. Scheidt and D. Pacico. Cooperative Localization and Mapping. Proceeding of the IEEE, Networking, Sensing and Control, pp. 596-601, 2006. [11] A. Marjovi, J. G. Nunes, L. Marques and A. de Almeida. Multi-robot exploration and re searching. Proceedings of the 2009 IEEE/RSJ international conference on Intelligent robots and systems, pp. 19291934, 2009. [12] A. Howard, M. J. Mataric and G. S. Sukhatme. Mobile Sensor Network Deployment using Potential Fields: A Distributed, Scalable Solution to the Area Coverage Problem. Proceeding of the 6th International Symposium on Distributed Autonomous Robotics Systems, pp. 299-308, 2002. [13] J. Tan, N. Xi, W. Sheng and J. Xiaov. Modeling multiple robot systems for area coverage and cooperation. Proceedings of the 2004 IEEE International Conference on Robotics and Automation, pp. 25682573, 2004. [14] Q. Jiang. An improved algorithm for coordination control of multiagent system based on r-limited voronoi partitions. Automation Science and Engineering, 2006. CASE 06. IEEE International Conference on, pp. 667-671, 2006. [15] F. WeiXing, W. KeJun, Y. XiuFen and G. ShuXiang, Novel Algorithms for Coordination of Underwater Swarm Robotics, Proc. of IEEE Int. Conf. on Mechatronics and Automation, pp 654-659, 2006. [16] R. A. Freeman, P. Yang and K. M. Lynch, Distributed estimation and control of swarm formation statistics, American Control Conf., 7 pp.-, 2006. [17] F. Kobayashi, N. Tomita, F. Kojima, Re-formation of mobile robots using genetic algorithm and reinforcement learning, SICE 2003 Annual Conf., vol. 3, pp. 2902-2907, 2003. [18] L. Barnes, W. Alvis, M. A. Fields, K. Valavanis and W. Moreno, Swarm Formation Control with Potential Fields Formed by Bivariate Normal Functions, 14th Mediterranean Conf. on Control and Automation, pp 1-7, 2006. [19] J. M. Hereford, A Distributed Particle Swarm Optimization Algorithm for Swarm Robotic Applications, Proc. of IEEE Congress on Evolutionary Computation (CEC), pp 1678-1685, 2006. [20] M. Hess, M. Saska and K. Schilling, Formation driving using particle swarm optimization and reactive obstacle avoidance, Proc. of 1st IFAC Workshop on Multivehicle Systems (MVS), pp 32-37, Lisboa, Portugal, 2006. [21] M. Dorigo, M. Birattari and T. Sttzle. Ant Colony Optimization Articial Ants as a Computational Intelligence Technique. IEEE Comput. Intell. Mag, pp. 28-39, 2006. [22] M. Dorigo. Optimization, learning and natural algorithms. PhD thesis, Dipartimento di Elettronica, Politecnico di Milano, 1992. [23] E. Bonabeau, M. Dorigo and G. Theraulaz. Inspiration of optmization from social insect behavior. Nature, pp. 39-42, 2000. [24] M. Dorigo, G. Di Caro and L. Gambardella. Ant Algorithms for Discrete Optimization. Technical Report IRIDIA/98-10, Universite Libre de Bruxelles, Belgium, 1999. [25] D. Yingying, H. Yan and Jiang Jingping. Multi-robot cooperation method based on the ant algorithm. Proceedings of the 2003 IEEE Swarm Intelligence Symposium, pp. 14-18, 2003. [26] B. Christian. Review of Ant colony optimization by M.Dorigo, T.Stützle, MIT Press, Cambridge, MA, 2004. Artif. Intell., pp. 261-264, 2005.

3255