Modern trends of artificial intelligence in humanitarian territory demining and it's integration into geoinformation systems

Abstract

Background. Humanitarian demining is distinguished by a variety of land-clearance methods, which differ in their physical principles of implementation. According to the readiness level (Technology Readiness Level – TRL), these technological solutions can be divided into routinely employed (TRL 8-9), systems tested in near-operational conditions (TRL 4-7), and conceptual technologies (TRL 1-3). None of the existing methods currently guarantees 100% effectiveness in humanitarian demining and are unsatisfactory in terms of sensitivity, selectivity and speed. The UN Mine Action Service requires mine action operators to achieve a near-to-full clearance confidence level of at least 99.5% down to a minimum depth of 20 cm. Demining activities are inherently geographical in nature, and therefore all spatially referenced data collected by various methods must be processed within Geographic Information Systems (GIS). The rapid expansion of artificial intelligence (AI) in recent years has demonstrated its ability to identify patterns and relationships in datasets far beyond human analytical capacity. The convergence of AI capabilities with geospatial data (GeoAI) has further enhanced the understanding of complex datasets, thereby contributing to solutions for some of humanity's most critical challenges, including that of a mine contamination.

Methods. Both general scientific and special research methods were applied. A historical approach allowed us to trace key stages of AI development. A classification method outlined the types of AI methods. A determination method defined the domain of AI application in humanitarian demining. Analysis and evolutionary approach enabled the examination of of AI application cases cross different stages of humanitarian demining cycle and their improvement through the development of geospatial AI. Induction and analysis formulated a conclusion regarding the current state, trends, and prospects of AI deployment in humanitarian land-demining.

Results. The study examines the historical emergence and growing importance of AI for the needs of humanitarian demining. A detailed review of publications on AI in humanitarian demining and geospatial AI was conducted. Trends and declared priorities for further development were identified. A classification of AI methods is provided with an explanation of their practical application in humanitarian demining. Neural networks are used as a case study to identify key challenges and outline feasible solutions for their integration into GIS workflows.

Conclusions. GIS and AI technologies are included in discussions on how technology and innovation can improve humanitarian and international peacekeeping activities. These technologies offer the potential for enhancing needs assessment, spatial change monitoring, and may benefit both the mine action sector and the broader humanitarian community.