In this article, a vision-based technique for obstacle avoidance and target identification is combined with haptic feedback to develop a new teleoperated navigation system for underactuated aerial vehicles in unknown environments. A three-dimensional (3-D) map of the surrounding environment is built by matching the keypoints among several images, which are acquired by an onboard camera and stored in a buffer together with the corresponding estimated odometry. Hence, based on the 3-D map, a visual identification algorithm is employed to localize both obstacles and the desired target to build a virtual field accordingly. A bilateral control system has been developed such that an operator can safely navigate in an unknown environment and perceive it by means of a vision-based haptic force-feedback device. Experimental tests in an indoor environment verify the effectiveness of the proposed teleoperated control.
Carloni, R., Lippiello, V., D'auria, M., Fumagalli, M., Mersha, A. Y., Stramigioli, S., & Sicilano, B. (2013). Robot vision: obstacle-avoidance techniques for unmanned aerial vehicles. Robotics and autonomous systems, 20(4), 22-31. https://doi.org/10.1109/MRA.2013.2283632