Active object recognition (AOR) aims at collecting additional information to improve recognition performance by purposefully adjusting the viewpoint of an agent. How to determine the next best viewpoint of the agent, i.e., viewpoint planning (VP), is a research focus. Most existing VP methods perform viewpoint exploration in the discrete viewpoint space, which have to sample viewpoint space and may bring in significant quantization error. To address this challenge, a continuous VP approach for AOR based on reinforcement learning is proposed. Specifically, we use two separate neural networks to model the VP policy as a parameterized Gaussian distribution and resort the proximal policy optimization framework to learn the policy. Furthermore, an adaptive entropy regularization based dynamic exploration scheme is presented to automatically adjust the viewpoint exploration ability in the learning process. To the end, experimental results on the public dataset GERMS well demonstrate the superiority of our proposed VP method.

The electromagnetic (EM) inversion refers to reconstructing the EM properties of unknown targets from measured scattered fields. In practical applications, various algorithms have been developed, especially those that rely on greedy pursuit (GP) within the compressive sensing framework. However, the GP-based algorithms still suffer from several intractable issues, such as poor priors of the pre-given sparsity index, insufficient robustness of generated single support, and possible failure when reconstructing non-sparse targets. A novel algorithm based on deep-first multipath matching pursuit (MMP-DF) assisted by a convolutional neural network (CNN) is proposed to overcome these issues. Specifically, the CNN is designed to predict the sparsity index, and MMP-DF is implemented to identify multiple promising supports for improving the likelihood of finding favorable support, and the Fourier basis is applied to transform the case of non-sparse targets into sparse ones. Theoretical analysis and numerical simulations show that, compared with the algorithm based on orthogonal matching pursuit (OMP), the proposed algorithm achieves improved and robust results in the case of non-sparse targets.

View planning (VP) is a technique that guides the adjustment of the sensor’s postures in multi-view perception tasks. It converts the perception process into active perception, which improves the intelligence and reduces the resource consumption of the robot. We propose a generic VP system for multiple kinds of visual perception. The VP system is built on the basis of the formal description of the visual task, and the next best view is calculated by the system. When dealing with a given visual task, we can simply update its description as the input of the VP system, and obtain the defined best view in real time. Formal description of the perception task includes the task’s status, the objects’ prior information library, the visual representation status and the optimization goal. The task’s status and the visual representation status are updated when data are received at a new view. If the task’s status has not reached its goal, candidate views are sorted based on the updated visual representation status, and the next best view that can minimize the entropy of the model space is chosen as the output of the VP system. Experiments of view planning for 3D recognition and reconstruction tasks are conducted, and the result shows that our algorithm has good performance on different tasks.