Multiagents Applied To Humanitarian Demining Pedro Santana1 , José Barata2 , and Luı́s Flores1 1 IntRoSys S.A. Quinta da Torre, Campus FCT-UNL 2829-516 - Portugal 2 New University of Lisbon, Quinta da Torre, Campus FCT-UNL 2829-516 - Portugal Abstract. The complexity associated to Humanitarian Demining becomes very high due to its broad set of activities, which beyond the already complex of landmine removal, includes other socio-economic supporting activities. Hence, more complex computer based supporting systems are required. The main goal of this article is to describe potential applications of multi-agent systems to the Humanitarian Demining domain, covering areas such as: knowledge-based systems, collaborative networks, agent-based modelling and multi-agent robotic systems. This is the result of the work being carried out by the Portuguese company IntRoSys, whose main research objective is the development of tools and methods to support humanitarian demining. 1 Introduction The complexity associated to Humanitarian Demining becomes very high due to its broad set of activities, which beyond the already complex of landmine removal, includes other socio-economic supporting activities. Hence, more complex computer based supporting systems are required, not only for the landmine removal itself, but for the socio-economic supporting activities. In addition to this these two different areas must be developed completely integrated, which just increases the complexity. Due to the involved complexity and the integration aspects that are required multiagents were considered as a suitable paradigm for supporting Humanitarian Demining. The main goal of this article is to describe potential applications of multiagent systems to the Humanitarian Demining domain. Section 2 introduces the Humanitarian Demining domain. Then, section 3 proposes a set of potential applications of multi-agents systems to Humanitarian Demining. Finally the conclusions are described in section 4. 2 Humanitarian Demining Concepts The estimated number for grounded anti-personal landmines all over the world is about 110 millions. Some consequences of using landmines are: unusable land (e.g. few landmines may hinder access to productive land), direct health consequences (e.g. amputation), indirect health consequences (e.g. victims’ families are entirely affected), and development consequences (e.g. communication paths closed). These reasons make urgent the development of new techniques that speed up and enhance the demining process. Humanitarian Demining is composed of two phases: the survey phase and the detection/clearance phase. Surveys are intended to rationalise the demining process, by analysing available resources, priorities, socio-economic impact, land end-use, etc. During the Impact Survey information to analyse priorities, social-economical impact, possible contaminated areas, etc. is gathered. Then, a Technical Survey is carried out to define the clearance requirements, the areas to be actually cleared, etc. Afterwards, the detection and clearance tasks are performed by following the requirements previously defined. Finally, a PostClearance Survey is performed to guarantee that the clearance requirements were met. 3 Multiagents Applied to Humanitarian Demining As previously stated, Humanitarian Demining covers areas from detection and clearance to the socio-economic impacts of such operations. Thus, the problem can be analysed in two different branches, the support that Multi-Agents can provide to minefield operations and to the process itself. 3.1 Operations Support It is well known the need for low-cost demining, so it can be afforded by local communities. Therefore, robotic systems must comply with such a requirement. Previous attempts to solve the problem of robotic demining as lead to high-cost solutions. Such attempts have tackled the problem with single robot systems, which carry all landmine detection sensors. Approaching the problem in a distributed perspective, where robots are simple and dispensable, allows to comply with the low-cost requirement. Hence, multi-agent systems, in particular multirobot systems, are a suited approach to the problem. Nevertheless, must not be forgotten that each robot is itself an agent with all the challenges behind this. We see with special interest Multi Unmanned Aerial Vehicles (UAV) that can be applied to assess the ground (e.g. using infra-red cameras) or just taking pictures faster and safer than ground vehicles. See for instance the ARC [4] project that makes use of airborne and satellite imaging to identify minefields. Resource configuration and allocation Multi-Agent based configuration architectures, such as the CoBASA architecture [2], can be well applied to prepare robotic teams for the minefield. It is important to integrate locals in the demining process, which are usually people with no formal education. Therefore, human-machine interfaces are both interesting and important to get closer high-tech and end-users. The user should interact with the system as it was actually in the minefield with its own methods. We envision a correct human-machine interface as the one capable of translating operational robotic semantics into operational end-user semantics, i.e. a correct ontological commitment. 3.2 Process Support Information has been identified as a crucial aspect in Humanitarian Demining. Humanitarian Demining is mainly about decision making, which relies heavily on data, information and knowledge. In this sense, Knowledge Based Systems are extremely relevant to this domain, mainly to support the decision in the impact study and mission configuration. The diversity of cultures and education level among the involved agents (e.g. locals, military, and governments) results in the need of intensive work on the area of Knowledge Engineering/Representation. To the problem of uncertain information is added the contradictions that may arise when merging different perspectives of the same problem. Hence, ontological commitments, belief revision and non-monotonic reasoning are certainly research opportunities for this domain. A global minefield atlas that aggregates digital topographic maps, satellite image data, demining related information and equipment suitability is presented in [7]. A Decision Support System (DSS) based on Geographical Information Systems (GIS) information and multi criteria analysis is proposed in [9]. The DSS integrates multi-layer information covering strategic level (e.g. economic value of the areas) down to operational level (e.g. demining company selection). Besides decision making intensive, Humanitarian Demining operations are highly distributed. Therefore, information sharing and collaborative tools are very important. Collaborative work tools may go from distributed devices that allow access to shared information sources (e.g. [6]) to autonomous mechanisms capable of managing the whole process. Humanitarian Demining could be modelled as a business composed of several parties. Following such a model, Collaborative Networks [3] as well as coalition formation [10, 1] and Multiagent Systems in general, may be interesting tools for the analysis and congregation of skills that are required to achieve the goal. Modelling the Humanitarian Demining process would allow to reason more effectively about it. In this sense, every entity could be modelled as an agent and the process as an Multi-Agent System. These models are extremely important, since decision making can rely upon scientific assessments reducing lobbying opportunities (i.e. objectiveness). Some tasks where agent-modelling and simulation could be applied are: to analyse relationships between innovations and performance gains, to optimise the selection and configuration of the elements for a demining campaign, to perform risk assessment by integrating models of the involved aspects, such as environment and local human activities. An approach for the economic modelling of cost-effective demining technologies is presented in ([11], pp. 19-26) and [5] presents a set of operational needs based on a demining model. 4 Conclusions A brief overview on the Humanitarian Demining domain was presented. It has been shown that the Humanitarian Demining domain is much broader than just demining itself, which opens new horizons for the application of technology on it. It has been proposed a set of scenarios where multi-agent system can be well applied, which includes: knowledge-based systems, collaborative networks, agent-based modelling and multiagent robotic systems. The Humanitarian Demining domain is a network of many heterogeneous entities (e.g. robots, decision makers, operators) with highly intricate connections. Thus, the Multi-Agents paradigm is well suited to model this distributed, heterogeneous, and complex structure. First preliminary results support this. References 1. Allsopp, D. N., Beautement, P., Kirton, M., Bradshaw, J. M., Suri, N., Knoblock, C. A., Tate, A. and Thompson, C. W.: Coalition Agents Experiment: Multiagent Cooperation in International Coalitions. IEEE Intelligent Systems, 173 (2002) 26– 35 2. Barata, J.: Coalition Based Approach for Shop Floor Agility - A Multiagent Approach. PhD thesis, Universidade Nova de Lisboa, Faculdade de Cincias e Tecnologia, Monte Caparica (2004) 3. Camarinha-Matos, L. M., Afsarmanesh, H. and Ollus, M.: Virtual Organizations Systems and Practices. Springer, New York (2005) 4. Eisl, M. and Khalili, M.: ARC – Airborne Minefield Area Reduction. In Proceedings of the International Conference Requirements and Technologies for the Detection, Removal and Neutralization of Landmines and UXO (2003) 5. GICHD/UNDP: Humanitarian Demining Equipment: Study of Global Operational Needs, Geneve (2002) 6. Horz, A. and Kunze, T.: MoMoSat – Collaborative Demining Information Management. In Proceedings of the Int. Conf. Requirements and Technologies for the Detection, Removal and Neutralization of Landmines and UXO (2003) 82–87 7. Littmann, F., Roux, S. and Sieber, A.: The global minefield atlas concept. In Proceedings of the Int. Conf. Requirements and Technologies for the Detection, Removal and Neutralization of Landmines and UXO (2003) 71–77 8. Long, M., Gage, A., Murphy, R. and Valavanis, K.: Application of the Distributed Field Robot Architecture to a Simulated Demining Task. In Proceedings of the IEEE International Conference on Robotics and Automation (2005) 9. Mladineo, N. and Knezic, S.: DSS for humanitarian Humanitarian Demining – case study Croatia. In Proceedings of the Int. Conf. Requirements and Technologies for the Detection, Removal and Neutralization of Landmines and UXO (2003) 93–98 10. Pechoucek, M., Marik, V. and Barata, J.: A Knowledge-based Approach to Coalition Formation. IEEE Intelligent Systems, 17 (2002) 17–25 11. Sahli, H., Bottoms, M. and Cornelis, J., eds: In Proceedings of the International Conference on Requirements and Technologies for the Detection, Removal and Neutralization of Landmines and UXO (2003) This article was processed using the LATEX macro package with LLNCS style View publication stats