Ground Impact and Hazard Mitigation for Safer UAV Operation

Authors: Andrew Poissant, Lina Castano, Huan Xu

Abstract

As Unmanned Aircraft Vehicles (UAV) become more commonplace, there is a growing need for safer flight software that allows the UAV to understand and autonomously react to numerous flight anomalies. Decision-making software must allow the aircraft to perform tasks such as detect and avoid obstacles, correct non-critical failures mid-flight, or select the safest response during the presence of a critical flight anomaly. This research, performed at the University of Maryland, develops a ground impact and hazard mitigation (GIHM) module that interfaces with nominal UAV flight control software, which ultimately allows for a UAV to choose the safest landing option in case of a critical flight anomaly. The module incorporates generating a feasible ground impact footprint in real-time with LandScan USA data, a dataset that contains accurate spatial population data with a resolution of 30 m. All model development and simulation was performed in Matlab and Simulink. 

Other works have used ground impact models for determining the reachable footprint for a UAV, and some have integrated census or tax data in a decision-making model that chooses the lowest hazard landing point. However, this work integrates a high-resolution population dataset and a decision-making engine with a model used to determine a UAV’s reachable footprint for many hazardous flight conditions. These conditions include loss of thrust and stuck ailerons, rudder, or elevator. Additionally, previous work did not integrate their ground impact models with flight control software, which is a fundamental feature of this work. The model we developed shows a preliminary reduction in simulated casualty expectation (fatalities per flight hour) by 97% when this new safety module is incorporated into a nominal UAV control software. This nominal flight software includes a mission plan, a path planning module, 6-DOF aircraft model, and flight controller. The path planning algorithm takes in the mission waypoints and real-time aircraft system states and updates the UAV’s flight plan. The updated flight plan is sent to the control system, where actuator commands for the aircraft are outputted and sent to the 6-DOF aircraft model. Finally, the 6-DOF aircraft states model outputs the real-time aircraft dynamics and sends those back to the path planning and control system blocks. The new ground impact hazard mitigation module takes in the aircraft dynamics, outputted by the 6-DOF aircraft states model, and uses those dynamics to determine whether the aircraft is experiencing a critical flight anomaly, calculates its reachable ground footprint based on the aircraft’s capabilities and flight dynamics, and determines the safest place to land based on expected population density in its potential landing areas. 

This work aligns with the “Trusting Autonomy” theme for HCSS because of the significant advances that have been made relating to safer autonomous control of UAVs. The poster details the architecture and development of the new safety module that reduces the casualty expectation of any UAV. Furthermore, this work aligns with the goals of HCSS because of its capability to ensure more dependable and safer UAV emergency landings.