Stereo-based path and obstacles detection in multiannual cultivations for autonomous navigation of an agricultural robot

ORELLANA, ADRIÁN; FERNANDEZ-LLORCA, DAVID; SORIA, CARLOS; LORENTE, ALEJANDRO

Proceedings of the 2012 IEEE Intelligent Vehicles Symposium Workshops

IEEE

Lugar: Alcalá de Henares; Año: 2012; p. 1 - 6

The intelligent agriculture is an area in hard andcontinuous development. This paper presents a new develop-ment carried out over an autonomous robot for agricultureenvironments to improve stereo vision processing. The auto-matic calibration system has been developed to obtain the pathplane and the path limits without the need of interaction witha human operator for a stereo camera mounted on the robot,with a pitch and roll not precisely defined previously. Then,the camera itself has been previously calibrated in order tobe able to give good 3D information referred to the cameracoordinates, but not to the world coordinates, depending on thevehicle position and the camera pitch and roll at every instant.The automatic calibration, working with the information thecamera collects from the environment, detects the path plane.From that information, generates an elevation map and from itdetects the path limits, helping the navigation system to adjustthe future position of the robot if necessary.