This Hovercraft is able to map its surroundings and to move directly in any given direction.
In the last few years, the exponential growth of the drone industry and the frequent use of such vehicles in the academic field have bring many new applications and innovations with their resulting consequences on the structural and software requirements of the embedded systems. Energy efficiency, weight gain and overall cost reduction are the main challenges and the most important factors affecting the actual capacity of a drone designed for autonomous indoor exploration to complete its tasks successfully. The ability to produce a cheap vehicle able to evolve in complex indoor environments at a big scale would help activities such as building surveillance, surface estimations, or military reconnaissance in dangerous areas.
In order to combine a better mobility while benefitting from the advantages of this kind of vehicle, we designed and implemented a remote controlled Hovercraft able to map its surroundings and to move directly in any given direction. The structure of the Hovercraft is based on five fans, four lateral fans are used to enable a direct movement in any direction while a central fan maintains an air cushion under the Hovercraft. This allows a directly multidirectional movement which can lead to a significant time gain in the case of indoor exploration. A turret equipped with two ultrasound sensors maps the perimeter around the hovercraft allowing an autonomous and safe exploration of the area. The resulting map is visualized by the user on a remote computer, allowing him to manually command the Hovercraft of to let him autonomously evolve in the area.
Hovercrafts can be an interesting alternative to flying drones since they can evolve on many types of environment while being able to navigate closely between obstacles. The principle of the Hovercraft being to maintain an air cushion on which the module can move with a very little amount of lateral power, it might be perfectly adequate for the kind of uses described before. Another challenge resides in the compromise between the complexity of the algorithms employed resulting in a hardware abundance and the quality of the data extracted from the system. While a sophisticated drone will be able to compute the information coming from numerous sensors and will react in real time to numerous unpredicted events, the loss in terms of battery life and extra cost generated can be superfluous depending on the situation. A little swarm of lighter Hovercrafts connected to a remote computer could be more cost efficient and obtain viable results way quicker than a single drone fully equipped.