Presently, multirotor unmanned aerial vehicles (UAV) are utilized in numerous applications. Their design governs the system’s controllability and operation performance by influencing the achievable forces and moments produced. However, unexpected causalities, such as actuator failure, adversely affect their controllability, which raises safety concerns about their service. On the other hand, their design flexibility allows further design optimization for various performance requirements, including actuator failure tolerance. Thus, this study proposed an optimization framework that can be employed to design a novel actuator fault-tolerant multirotor UAV configuration. The approach used an attainable moment set (AMS) to evaluate the achievable moment from a multirotor configuration; similarly, standard deviation geometries (SDG) were employed to define performance requirements. Therefore, given a UAV configuration, actuator fault situation, and SDG derived from the designed mission requirement, the suggested optimization framework maximizes the scaling factor of SDG and fits it into the AMS by adjusting the design parameters up to a sufficient margin. The framework is implemented to optimize selected parameters of the Hexacopter-type of parcel delivery multirotor UAV developed by the PNU drone, and a simulation was conducted. The result showed that the optimized configuration of the UAV achieved actuator fault tolerance and operation-performing capability in the presence of a failed actuator.
In this study, an improved model of the electric propulsion system is proposed in order to analyze the performance of large-size multicopter unmanned aerial vehicles. The main improvement of the proposed model is to reflect the armature reaction of the motor, which effectively explains the significant performance degradation in high-power operation. The armature reaction is a phenomenon, in which the main field flux is interfered by a magnetic flux and, as the size and output of the motor increase, the effect of armature reaction also rapidly increases. Therefore, the armature reaction must be considered for the optimal design and performance analysis of large-size multicopter platforms. The model proposed in this study includes several mathematical models for propellers, motors, electric speed controllers, and batteries, which are key components of the electric propulsion system, and they can calculate key performance data, such as thrust and torque and power consumption, according to given product specifications and input conditions. However, estimates of the armature reaction constants and heat profiles of motors need to be obtained in advance through experimental methods, since there is not yet enough data available in order to derive an estimation model. In conclusion, a comparison with the static thrust test of some commercial products confirmed that the proposed model could predict performance in the high-power operation of electric propulsion systems for large multicopter platforms, although some errors were noted.
Unmanned aerial vehicles (UAVs) are being widely utilized for various missions: in both civilian and military sectors. Many of these missions demand UAVs to acquire artificial intelligence about the environments they are navigating in. This perception can be realized by training a computing machine to classify objects in the environment. One of the well known machine training approaches is supervised deep learning, which enables a machine to classify objects. However, supervised deep learning comes with huge sacrifice in terms of time and computational resources. Collecting big input data, pre-training processes, such as labeling training data, and the need for a high performance computer for training are some of the challenges that supervised deep learning poses. To address these setbacks, this study proposes mission specific input data augmentation techniques and the design of light-weight deep neural network architecture that is capable of real-time object classification. Semi-direct visual odometry (SVO) data of augmented images are used to train the network for object classification. Ten classes of 10,000 different images in each class were used as input data where 80% were for training the network and the remaining 20% were used for network validation. For the optimization of the designed deep neural network, a sequential gradient descent algorithm was implemented. This algorithm has the advantage of handling redundancy in the data more efficiently than other algorithms.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.