Virtual Cage for Cost-Effective MAV Systems and Algorithms Prototyping

Transitioning from a simulation environment to real flight can be costly for MAV system and algorithm designers as they wrestle with the challenges and unpredictability of the real world. This transition often involves problems that can result in hardware damage or possibly injury. We have developed a system that cost-effectively addresses these problems. The system creates a virtual cage in which the MAV flight parameters are constrained to ensure the detection of the virtual cage boundaries and the safe recovery when those boundaries are reached. The system can operate with a minimal infrastructure and on-board sensing capabilities, functions indoors and outdoors, supports a range of MAVs including those with low computational capabilities, and automatically identifies the constraints to impose on the MAV based on its sensing and navigation capabilities and the cage attributes. The empirical assessment of the system with a MAV colliding against the virtual boundaries nearly 250 times in indoor and outdoor settings showed that the MAV remained within the virtual bounds over 97% of the time.