Dear Colleague,

The Neural, Emergent and Agent Technology (NEAT) group at the Department of Computer Science, University of Hull, UK, invites you to participate in a network of excellence preliminarily titled "The HumanITarian Demining Project". The network will apply for funds under the EU 6^th Framework Programme for Research and Technological development in late 2002, but will be submitting an Expression of Interest (EoI). Included below is a preliminary brief description of the focus of the network. If you are interested in participating, you are invited to propose changes and additions to the text, both for the EoI and later for the network formation. Declarations of interest in the present stage are not binding with respect to the application in late autumn this year. To be able to submit the EoI, we need (1) to know if you are interested in participating and (2) your suggestions about the content before May 15^th.

Anti-Personal landmines are a significant barrier to economic and social development in a number of countries, especially Vietnam, Bosnia & Herzogovina, Croatia, Serbia, Cambodia, Afghanistan and Angola. Since the 1940's, metal detectors have been used to detect buried landmines and although several other methods have been studied and used, this method remains unchanged today. The underlying problem with landmine research is in fact not the actual technology being used but rather the way in which research is conducted. The landmine research community lacks communication and coherence and this has consequently slowed down the demining process. Landmine detection is a subject that requires an integration of knowledge from different areas of expertise.

The network will contribute to an International European research environment by integrating the practical knowledge of the demining personnel with the theoretical knowledge of the sensor experts and engineers.

The most important area is that of the deminer who has expert knowledge in minefields and practical problems involved in mine clearance. Sensor experts who are familiar with the technology currently being used for military demining and humanitarian demining also play an important role in research of landmine detection. Additionally, two relatively new areas need to be explored further: the role of the autonomous robot in detection and clearance and the role of intelligent fusion methods of existing sensor technologies. The above areas must be integrated for any sort of progress in Humanitarian demining.

The network’s contribution in this respect will be twofold: to integrate existing experts and their valuable knowledge in humanitarian demining research and to open doors for new research in landmine detection and clearance.

Humanitarian Demining consists of both mine detection and mine clearance. This includes defining potential areas contaminated by mines and unexploded ordnance, preparing the terrain for the areas, detecting the mines and clearing them. Once this is accomplished, the cleared areas are assessed and further mine detection and clearance is performed if necessary. The most difficult step is the detection of mines and the following two ideas are proposed:

The most common form of mine detection is through hand held metal detectors and trained sniffer dogs. Unfortunately, this can be an extremely tedious and slow process. Metal detectors currently used by demining teams cannot differentiate a mine from metallic debris, which leads to 100-1000 false alarms for each real mine in minefields where the soil is contaminated by large quantities of shrapnel, metal scraps and cartridge cases.

All efforts have been directed towards an improved mine detector but what is essentially needed is a detector/sensor that will reliably confirm that the ground being tested does not contain an explosive device, with a reliability approaching 100%. There are undoubtedly a large number of sensors being currently investigated for the purpose of landmine detection but none of these technologies presented seems in fact capable of reaching, in a very large number of situations, good enough detection while maintaining a low false alarm rate. Rather, each one will probably have to find, if it exists, a specific area of applicability, determined by technological as well as economical or even social factors, and possibly other sensors to work with using some form of sensor fusion.

Sensor fusion using soft computing methods such as fuzzy logic, neural networks and rough set theory must be further explored and a computationally inexpensive method of combining sensory data must be designed. These methods should also have the capability to assess the quality of the mined area once the mines have been cleared.

Another existing problem is testing. Most of the research produced by sensor fusion experts have only been tested on a test bed - i.e. an environment were buried landmines have the explosives removed before testing. Without the metal encased detonators, mines are much more difficult to detect than they would be under operational conditions. So when a system cannot pick up the mine - it is incorrectly classified as unsuitable or fine tuned so much that it becomes far too sensitive. On-site experiments and testing is extremely vital and this is something we hope will be solved by bridging the communication gap between sensor experts and deminers.

There is still a large amount of scepticism on the use of autonomous robots for demining purposes and this stems from the lack of communication between the experts in the robotics field and the demining experts. The general consensus on robots in the minefield is that (1) the deming problem is more sensor related than platform related and (2) robots are too expensive to maintain and use by the host country.

The first point is an extremely vital one: just fancy mobile robotics or any other kind of unmanned platform will not solve the landmine problem. On the other hand, it has been hypothesised that a mobile platform with the right kind of sensors is more capable of dealing with the detection of landmines than an intelligent sensor on a hand held device. A team of robots employed to demine a specific area of terrain using the intelligent sensor fusion techniques that we are currently developing will improve the efficiency and speed of target acquisition in mined areas. Additionally, the use of robots reduces the role of the human on the actual mine field.

There is still the issue of expense and finding a cost effective solution. The host countries that are affected by landmines cannot afford expensive equipment such as intelligent mobile robotics. Again, this is a problem that can only be solved through time and the rapid advance of technology. Building cheap and simple robots is much easier than it was a decade ago. The robot must not only be cheap to build but easy to maintain and program for a novice in robotics such as a deminer from the host country.

Communication remains a major issues to be solved if the objectives above are to be met. So far, the expert in robotics knows too little about the practical challenge of demining: hence the robot is designed like all other autonomous robots attempting to navigate an unknown environment. Although some aspects of navigation may be extended to demining robots, it will be more reliable if robots were designed specifically for the purpose of landmine detection than as an after thought.

On the other side, the deminer knows too little about robots to allow new ideas for deming robots to filter into the scientist’s working area and sceptism grows.

It is important to note that looking for a huge breakthrough using sensor fusion and robotics would only lead to more expensive toys that cannot be used in areas that matter. On the other hand, improvement by small innovations may eventually provide an answer to the crucial problem of landmine detection. This network of excellence hopes to bring together people of different expertise and thus make it possible for improvements in landmine detection and clearance.

For further information and interest please contact:

Dr S.Rajasekharan