Deep Learning Quadcopter Control via Risk-Aware Active Learning

Modern optimization-based approaches to control increasingly allow automatic generation of complex behavior from only a model and an objective. Recent years has seen growing interest in fast solver ...