Risk Assessment, Prediction, and Avoidance of Collision in Autonomous Drones

Anamta Khan
DOI: https://doi.org/10.48550/arXiv.2108.12770
IF: 3.7
2021-08-29
Robotics
Abstract:Unmanned Aerial Vehicles (UAVs), in particular Drones, have gained significant importance in diverse sectors, mainly military uses. Recently, we can see a growth in acceptance of autonomous UAVs in civilian spaces as well. However, there is still a long way to go before drones are capable enough to be safely used without human surveillance. A lot of subsystems and components are involved in taking care of position estimation, route planning, software/data security, and collision avoidance to have autonomous drones that fly in civilian spaces without being harmful to themselves, other UAVs, environment, or humans. The ultimate goal of this research is to advance collision avoidance and mitigation techniques through quantitative safety risk assessment. To this end, it is required to identify the most relevant faults/failures/threats that can happen during a drone's flight/mission. The analysis of historical data is also a relevant instrument to help to characterize the most frequent and relevant issues in UAV systems, which may cause safety hazards. Then we need to estimate their impact quantitatively, by using fault injection techniques. Knowing the growing interests in UAVs and their huge potential for future commercial applications, the expected outcome of this work will be helpful to researchers for future related research studies. Furthermore, we envisage the utilization of expected results by companies to develop safer drone applications, and by air traffic controllers for building failure prediction and collision avoidance solutions.
What problem does this paper attempt to address?