A scientist created drones able to learn new routes while flying autonomously. The robot drones will not need on ground pilots. These aerial vehicles are able of autonomous flight, using a vision and learning system that control and navigate them without relying on trained personnel or a GPS signal.
The researcher from the National Institute of Astrophysics, Optics and Electronics in Mexico, José Martínez Carranza, designed the vision and learning system for flight control of the autonomous drones.
The Mexican researcher developed an innovative method for estimating the orientation and position of the aerial vehicle. The on-board system is allowing the drone to recognize its environment, replacing the GPS location system with some low-cost sensors such as camcorders, gyroscopes and accelerometers.
The aim of Carranza's system was to use on-board video cameras and avoid the use of GPS. His control system is using the visual information from the cameras in tandem with an algorithm to locate and orient the drone during its flight. For this purpose, the researcher adapted a function that allows using aerial view for drawing a specific route on a map. The system is similar in a way to Google maps, indicating the route to a particular destination for autonomous navigation.
The knowledge used in this project was developed in the research of UAVs' precise navigation in complex environments. José Martínez Carranza developed this study during his postdoc at the University of Bristol. The British company Blue Bear Ltd. took part in the project as well, providing the control algorithms and the drones. The financing for this project was obtained from the Defense Science and Technology Laboratory, Innovative U.K. and several British government agencies specialized in financing technological innovation projects.
Carranza returned to Mexico upon completion of these projects, working as full-time researcher at the INAO. There, the Mexican scientist has won the British Academy of Sciences' financing award Royal Society-Newton Advanced Fellowship. The financing has been used by Carranza to perform science research focused on aerial robotics, according to the researcher. José Martínez Carranza also holds a Computer Science Master degree.
He obtained funding for a project on UAVs autonomous flight in GPS-denied outdoor areas, with the objective to investigate methods to perform drone autonomous flight on environments with limited computational processing capabilities, no GPS signal and challenges as wind currents.
The system developed by Carranza requires the intervention of a pilot only when the drone takes off. Once in the air the algorithms of autonomous flight come into action and the aerial vehicle processes visual information captured by the camera in order to recognize its position and navigate to each of the points made in the route recorded.
