Conceitos essenciais
The author proposes a perceptive safety filter integrating Control Barrier Functions with VI-SLAM and dense 3D mapping for safe teleoperated MAV navigation in complex environments.
Resumo
The content discusses a system that ensures safe navigation of Micro Aerial Vehicles (MAVs) in complex environments through a perceptive safety filter. The system combines Control Barrier Functions, Visual-Inertial Simultaneous Localization and Mapping (VI-SLAM), and dense 3D occupancy mapping to guarantee safe navigation. The approach closes the perception-action loop, demonstrating the full capability of safe control without external infrastructure or prior knowledge of the environment. Experimental results from simulations and real-world scenarios validate the efficacy of the proposed system.
Key points include:
- Introduction of a perceptive safety filter for teleoperated MAVs.
- Integration of Control Barrier Functions, VI-SLAM, and dense 3D mapping.
- Real-time updates based on onboard sensors and computation.
- Successful demonstration in simulated and real-world experiments.
Estatísticas
"We define a point in 3D space as unsafe if it satisfies either of two conditions: (i) it is occupied by an obstacle, or (ii) it remains unmapped."
"Our system relies solely on onboard IMU measurements, stereo infrared images, and depth images."
"In both cases, a map of the MAV’s environment is required so that regions where it is safe to fly can be determined."
Citações
"Our system relies solely on onboard IMU measurements, stereo infrared images, and depth images."
"In contrast to existing perception-based safe control frameworks, we directly close the perception-action loop."