This academic paper addresses the challenges associated with trajectory planning for affordable and light-weight Unmanned Aerial Vehicle (UAV) swarms, despite limited computing resources and extensive cooperation requirements. Specifically, an imitation-based starling cluster cooperative trajectory planning technique is proposed for a fixed-wing model of a six-degree-of-freedom UAV cluster. To achieve this, dynamic trajectory prediction of the rapid random search tree is utilized to generate a track solution adapted to the terrain environment. Additionally, the Dubins aircraft path solution is applied as it is suitable for executing input track commands by the UAV model. Computational simulations on different cluster sizes show the approach can maintain the cluster state while navigating diverse terrains, with the track solution complying with the UAV’s physical model properties.

Cooperative formation control of unmanned ground vehicles (UGVs) has become one of the important research hotspots in the application of UGV and attracted more and more attention in the military and civil fields. Compared with traditional formation control algorithms, reinforcement-learning-based algorithms can provide a new solution with a lower complexity for real-time formation control by equipping UGVs with artificial intelligence. Therefore, in this paper, a distributed deep-reinforcement-learning-based cooperative formation control algorithm is proposed to solve the navigation, maintenance, and obstacle avoidance tasks of UGV formations. More importantly, the hierarchical triangular formation structure and the newly designed Markov decision process for UGV formations of leader and follower attributes make the control strategy learned by the algorithm reusable, so that the formation can arbitrarily increase the number of UGVs and realize a more flexible expansion. The effectiveness and scalability of the algorithm is verified by formation simulation experiments of different scales.

Many unsupervised domain adaptation methods require simultaneous access to both source-domain and target-domain data during the training stage. However, in practical electro-mechanical actuator (EMA) fault diagnosis, source-domain data may be unavailable, or it might not be able to be directly shared because of data ownership, confidentiality, or deployment constraints. Moreover, variations in command waveform, amplitude, load, and operating speed can cause distribution shifts between different working conditions, which degrades the diagnostic performance of models trained under a single condition. The domain adaptation model will also experience decreased performance in the source domain due to catastrophic forgetting. Hence, a dual-domain adaptation model is proposed, one which can reduce the risk of data leakage and relieve catastrophic forgetting in the source domain while improving domain adaptation in the target domain. The proposed framework transfers a source-trained model to the target domain without directly accessing source-domain samples. Sparse domain attention is introduced into the feature network to construct source-related and target-related feature channels, aiming to reduce source-domain performance degradation during target-domain adaptation. In addition, the classifier is fixed according to the source-domain hypothesis, and local structure clustering is used to improve pseudo-label consistency for unlabeled target-domain samples. The effectiveness of the proposed method is compared with several methods, utilizing low-frequency and high-frequency monitoring signals of the electro-mechanical actuator dataset, respectively.