Project Description

Most drones today come with a built-in GPS system. This allows the drone to fly back to "base" if the batteries are low or if the flyer loses sight of the drone. But sometimes, there may be GPS reception problems, especially in big, open areas. This can happen if the signals are disturbed by cables underground or the area has other geographical problems, like big rocks or walls. Unfortunately, this is simply a problem with the location. The GPS system is not available indoors so none of the drones depend on navigation will not work inside a building or an office. So if this happens, the drone flying should be taken elsewhere to a place that receives better GPS which is not possible in indoor system or else drones must be developed without using GPS system. The optimal solution would be developing a drone without GPS system. This can be achieved when the drone is provided with intelligence. The context in this project is autonomous navigation of drones both indoor and outdoor. The outdoor applications of this project involves search for survivors in damaged buildings and inspect railroad tracks in tunnels. The indoor application involves Indoor Surveillance and check stock on store shelves. Development of AI-powered drone involves four important technologies drone development; Image processing, Digital Twin and Indoor positioning. Using these four technologies the both indoor and outdoor autonomous navigation of AI powered drone is achieved. On board computation is very crucial to apply neural networks so NVidia Jetson was choose as the master board for this project. The whole core of the project is to develop a drone that navigates without using GPS. Numerous applications can be extended like detecting forest fires, Track a person in forest and track number of trees in forest. These applications prove the potential of the project and this is going to create a base for autonomous drones in India. This is going to transform the ideology and creates a social impact on drones in India. Many rescue operations can be performed and this technology is very helpful for defence and military operations. This project proposes using a Drone to produce a solution to an autonomous search robot. Equipped with extra sensors and using AI or agent based programming the drone will be required to roam an indoor environment searching for colours, objects, and people. The project will look into the scope of what objects can easily be detected using the on board camera, as well as other on board sensors, and which image processing algorithms are best for doing this both accurately and quickly on a moving platform. The project also covers looking into navigating an unknown indoor environment using search algorithms, testing and which of a variety of algorithms work best for a hovering UAV. Eventually the drone should be able to given an image of an object or person to find and perform autonomous actions to locate that object or person, without having an altercation with the environment around it. The same context can be applied outdoors irrespective of the type of environment around the drone.