2025 11 13 pathfinder

Pathfinder

CIHBw-4 Bomb-Lense

Problem - Development of an app that enables enemy weapon systems to be detected, evaluated and shared in the context of a unified situational picture. The app is designed to scan objects such as crashed drones, ammunition parts and unexploded ordnance and identify them automatically using AI. Based on the identification, relevant information about the object, such as the danger radius and the potential risk, is shared with the soldiers.

Context Information about individual weapon systems can influence the course of battles and the weighing up of options for action. At the same time, there are more and more opportunities to perform AI-based image classification close to the edge. An app that can be widely installed on existing mobile phones and can cope with network restrictions at the edge is therefore helpful.

Scenario A soldier finds an unexploded bomb near the front line and wants to know whether it was fired by his own troops or by the enemy. He wants to know what danger he might be exposed to if more ordnance of this type arrives in his area of operation.

  • Use

    • Android Mobile App : https://github.com/dwani-ai/dwani-android
    • Server : https://github.com/dwani-ai/dwani-api-server

  • Previous Work

    • Ghost Vision - https://github.com/dwani-ai/llm-recipes/tree/main/python/ghost-vision
    • Recon - https://github.com/dwani-ai/llm-recipes/tree/main/python/reconaissance
    • Aquila - https://github.com/dwani-ai/llm-recipes/tree/main/python/aquila