Efficient Generation of Ground Impact Probability Maps by Neural Networks for Risk Analysis of UAV Missions

Abstract

This paper investigates the generation of ground impact probability maps for UAVs in case of failure during the flight. Such maps are of a huge interest for risk assessment of UAV operations and can be used both for offline mission preparation or analysis and online decision making. Two approaches are proposed in this paper to generate such maps, taking into account a dynamical model a fixed-wing UAV and wind conditions. The first one relies on the generation of a complete database by Monte Carlo simulations. The second one is based on neural network surrogate models obtained by supervised learning using this database. Computation time required by the second approach is very small and compatible with online use. The two approaches are presented and discussed, and examples of ground impact probability maps generated are provided.

DROSERA: A DROne Simulation Environment for Risk Assessment

“…This solution enables fast risk computation by the DROSERA tool, while preserving accuracy compared to a simplistic model. Note that, development of surrogate models based on neural networks (Levasseur et al (2020)) are also under investigation for computation time reduction.…”

Risk-Aware Guidance of a Fixed-Wing UAV using Neural Network Model Predictive Control

“…Probability of the risk for people at ground is accounted for in the formulation of the MPC problem. Computation of predictions of this risk probability is done online by using neural networks designed in previous works by the authors [13] [18]. More precisely, these neural networks enable accurate and fast computation of the probability of ground impact by also taking into account the influence of altitude and speed of the UAV, as well as wind direction and speed.…”

Drone Ground Impact Footprints with Importance Sampling: Estimation and Sensitivity Analysis

“…Total loss of power is assumed and uncertainties on the initial conditions of the UAV at failure instant as well as on the deflection of unactuated control surfaces are considered. A 6 degreesof-freedom flight mechanics model is also used in [11] for a fixed wing UAV to estimate ground impact probability maps, taking into account the influence of wind direction and speed. Real flight data have been used to model uncertainties on the turn rate and flight path angle of the vehicles for cruise-like mode at constant altitude and straight line.…”

“…K-Nearest neighbors models have been considered in [10] to approximate impact probability distribution. Other techniques such as Krigging have been investigated [7] regarding impact footprints or neural networks for both generation of impact footprints [7] and probability maps [11].…”