Risk-Averse Motion Planning for Quadrotors Using Conditional Value-at-Risk

This article presents a risk-averse motion planning method with conditional value-at-risk (CVaR) for the safe operation of quadrotors subject to multiple uncertainties. Specifically, the position uncertainty of the center of mass (CoM) of the obstacle and the battery power consumption uncertainty, which lead to the collision risk and the energy risk, respectively, are considered. The two types of uncertainties are depicted by their respective stochastic distributions. The Copula function is used to capture the statistical correlation between stochastic position coordinates. In order to quantitatively characterize the collision risk and the energy risk, a CVaR-based risk measure method is developed to map the stochastic distributions to specific quantities. Combining these two risks, the dual-risk cost function is formulated and optimized in the trajectory planning layer to generate the minimum risk trajectory. Experimental results demonstrate the feasibility and effectiveness of the proposed risk-averse motion planning method.