The view planning (VP) problem in robotic active vision enables a robot system to automatically perform object reconstruction tasks. Lacking prior knowledge, next-best-view (NBV) methods are typically used to plan a view sequence, with the goal of covering as many object surface areas as possible in an unknown environment. However, such methods have two problems: (1) they are unable to perform global path planning; and (2) the reconstruction process is inefficient because of time-consuming ray casting and high movement cost. We propose a neural network, SCVP, to pre-learn prior knowledge via set covering (SC) based training so as to achieve one-shot view planning. The SCVP network takes the volumetric occupancy grid as input and directly predicts a small (ideally minimum) number of views that cover all surface areas. Given object 3D models as a priori geometric knowledge, the training dataset is automatically labeled by the set covering optimization method. We propose a global path planning method to reconstruct objects without redundant movement. Comparative experiments on multiple datasets of 3D models show that, under the condition of similar or better surface coverage, the proposed method can outperform state-of-the-art NBV methods in terms of movement cost and inference time. Real-world experiments confirm that the proposed method can achieve faster object reconstruction than other methods.

Helping robots understand indoor scenes has considerable value in computer vision. However, due to the diversity of indoor scenes, understanding them remains a big challenge. There are many spatial right-angles in indoor scenes. These spatial right-angles are projected into diverse 2D projections. These projections can be considered a composition of a pair of lines (line-pairs). Given the vanishing points (VPs), line segments can be assigned to 1 of 3 main orthogonal directions. The line-pairs (intersection of 2 lines), such that each of them converges to a different VP, are likely to be the projection of a spatial right-angle onto the image plane. These projections may enable us to estimate their original orientation and position in 3D scenes. In this paper, we presented a method to efficiently understand indoor scenes from a single image, without training or any knowledge of the camera's internal calibration. Through geometric inference of line-pairs, it is possible to find these spatial right-angle projections. Then, these projections can be assigned to different clusters, and the line that lies in the neighbor-cluster helps us estimate the layout of the indoor scene. The proposed approach required no prior training. We compared the room layout estimated by our algorithm against the room box ground truth, measuring the percentage of pixels that were correctly classified. These experiments showed that our method estimated not only room layout, but also details of the indoor scene.