For the problem of gait stability control for continuous linear walking of a biped robot, a Soft Actor-Critic (SAC) gait control algorithm based on maximum entropy Deep Reinforcement Learning (DRL) was proposed. Firstly, without accurate robot dynamic model built in advance, all parameters were derived from joint angles without additional sensors. Secondly, the cosine similarity method was used to classify experience samples and optimize the experience replay mechanism. Finally, reward functions were designed based on knowledge and experience to enable the biped robot continuously adjust its attitude during the linear walking training process, and the reward functions ensured the robustness of straight walking. The proposed method was compared with other DRL methods such as PPO (Proximal Policy Optimization) and TRPO (Trust Region Policy Optimization) in Roboschool simulation environment. The results show that the proposed method not only achieves fast and stable linear walking of the biped robot, but also has better algorithmic robustness.
Multi-task Joint Learning Model (MJLM) was proposed to solve the performance improvement bottleneck problem caused by the separation of viewpoint-invariant feature and view transformation method in the existing cross-view geo-localization methods. MJLM was made up of a proactive image generative model and a posterior image retrieval model. In the proactive generative model, firstly, Inverse Perspective Mapping (IPM) for coordinate transformation was used to explicitly bridge the spatial domain difference so that the spatial geometric features of the projected image and the real satellite image were approximately the same. Then, the proposed Cross-View Generative Adversarial Network (CVGAN) was used to match and restore the image contents and textures at a fine-grained level implicitly and synthesize smoother and more real satellite images. The posterior retrieval model was composed of Multi-view and Multi-supervision Network (MMNet), which could perform image retrieval tasks with multi-scale features and multi-supervised learning. Experimental results on Unmanned Aerial Vehicle (UAV) dataset University-1652 show that MJLM achieves the Average Precision (AP) of 89.22% and Recall (R@1) of 87.54%, respectively. Compared with LPN (Local Pattern Network) and MSBA (MultiScale Block Attention), MJLM has the R@1 improved by 15.29% and 1.07% respectively. It can be seen that MJLM processes the cross-view image synthesis and retrieval tasks together to realize the fusion of view transformation and viewpoint-invariant feature methods in an aggregation, improves the precision and robustness of cross-view geo-localization significantly and verifies the feasibility of the UAV localization.
Aiming at the problem of low accuracy of the existing cross-view image matching algorithms, an Unmanned Aerial Vehicle (UAV) image localization method based on Multi-view and Multi-supervision Network (MMNet) was proposed. Firstly, in the proposed method, satellite perspective and UAV perspective were integrated, global and local features were learnt under a unified network architecture, then classification network was trained and metric tasks were performed in multi-supervision way. Specifically, the Reweighted Regularization Triplet loss (RRT) was mainly used by MMNet to learn global features. In this loss, the reweighting and distance regularization strategies were to solve the problems of imbalance of multi-view samples and structure disorder of the feature space. Simultaneously, in order to pay attention to the context information of the central building in target location, the local features were obtained by MMNet via square ring cutting. After that, the cross entropy loss and RRT were used to perform classification and metric tasks respectively. Finally, the global and local features were aggregated by using a weighted strategy to present target location images. MMNet achieved Recall@1 (R@1) of 83.97% and Average Precision (AP) of 86.96% in UAV localization tasks on the currently popular UAV dataset University-1652. Experimental results show that MMNet significantly improves the accuracy of cross-view image matching, and then enhances the practicability of UAV image localization compared with LCM (cross-view Matching based on Location Classification), SFPN (Salient Feature Partition Network) and other methods.
Focusing on the higher ratio of processor utilization and lower execution cost of a scientific workflow in cloud, a policy of execution optimization based on task cluster aggregation was proposed. First, the tasks were reasonably replicated and aggregated into several clusters. Therefore, the key tasks could be scheduled as early as possible. Then, the task clusters were aggregated again to facilitate the spare time among the tasks in the task cluster. The experimental results show that the proposed policy can improve the parallelism of workflow tasks, advance the earliest finish time of the whole workflow and it has a significant effect in improving the utilization ratio of processors and lowering the cost of workflow execution.
Semi-Supervised Support Vector Machine using label mean (meanS3VM) for image classification selects a small number of unlabeled instances randomly to train the classifier, and the classification accuracy is low; meanwhile, the parameter's determination always derives much oscillation of the results. In allusion to the above problems, meanS3VM image classification method based on mean shift was proposed. The smoothed image acquired by mean shift was used as original segmented image to reduce diversities of image features; an instance in each smoothed area was randomly selected as unlabeled instance to ensure that it carried useful information for classification and had a more efficient classifier; and the parameters value were also investigated and improved, the grid search method was used for sensitive parameters, the parameter ep was estimated by combining with Support Vector Machine (SVM) mean shift results, so that there will be a better and more stable result. The experimental results indicate that the classification rate of the proposed method to ordinary and noise image can be averagely increased more than 1% and 5%, and it has higher efficiency and avoids the oscillation of the results effectively, which is suitable for image classification.