Advancements in Autonomous Aerial Navigation

The field of autonomous aerial navigation is moving towards the development of more resilient and safe control systems. Researchers are exploring the combination of learning-based and safety controllers to improve the navigation of quadrotors in complex environments. The use of probabilistic risk assessment and reachability analysis is also becoming increasingly popular for designing risk-bounded controllers. Additionally, reinforcement learning with privileged information is being used to navigate around large obstacles. Noteworthy papers include: Improving the Resilience of Quadrotors in Underground Environments by Combining Learning-based and Safety Controllers, which proposes a combined controller that switches between learning-based and safety controllers. PRREACH: Probabilistic Risk Assessment Using Reachability for UAV Control, which presents a new approach for designing risk-bounded controllers using reachability analysis. Quadrotor Navigation using Reinforcement Learning with Privileged Information, which leverages privileged information to navigate around large obstacles.

Sources

Improving the Resilience of Quadrotors in Underground Environments by Combining Learning-based and Safety Controllers

Hierarchical Low-Altitude Wireless Network Empowered Air Traffic Management

PRREACH: Probabilistic Risk Assessment Using Reachability for UAV Control

Quadrotor Navigation using Reinforcement Learning with Privileged Information

Built with on top of