A novel low-altitude reconnaissance strategy for smart UAVs: active perception and chaotic navigation

Abstract

In this article, an innovative approach is proposed for multi-sensory integration applied in autonomous navigation for smart low-altitude reconnaissance unmanned aerial vehicles (UAVs). The new UAV navigation technique exploits the use of a pan—tilt-based visual sensing system, which is able to implement the strategy of active perception, increasing the richness of the information acquired from the environments. In addition, based on the information theory, a significant optimality criterion, so-called observation entropy, is represented to determine which set of landmarks should be adopted as the optimal set of information source. Moreover, a chaotic reaction system, inspired by the olfactory bulb neural activity observed in rabbits subject to external stimuli, is considered to build the perception-motion dynamics loop. The loop can be used to adaptively guide the UAV actions not only when it perceives a set of landmarks, but also when the UAV is in environments without enough landmarks. Finally, the simulations and computed results show that the introduced autonomous navigation strategy has very low error rates of the path estimation supported by enough landmarks and active perceptual actions. Furthermore, the results also indicate that the chaotic evolution-based strategy is provided with robustness, which can guarantee the continuous navigation for the UAV in the complex environments without enough landmarks.