Sensory predictive guidance in partially known environment

This paper addresses the problem of autonomous navigation through a partially known cluttered environment. The proposed hierarchical framework is developed as an extension to the recently proposed guidance algorithm based on Receding Horizon optimization with Spatial Value or Cost-to-Go function. It ensures a tight integration between the environment map update (via an on-board depth sensor), cost-to-go update as well as the low-level control system. The overall approach combines key elements from robotic motion planning and trajectory optimization and addresses the particular challenges posed by dynamical systems in partially known environment. A simulation example with a Blade-Cx2 indoor helicopter is used to demonstrate the proposed guidance framework.