No abstract

We study the problem of extracting correspondences between a pair of point clouds for registration. For correspondence retrieval, existing works benefit from matching sparse keypoints detected from dense points but usually struggle to guarantee their repeatability. To address this issue, we present CoFiNet -Coarse-to-Fine Network which extracts hierarchical correspondences from coarse to fine without keypoint detection. On a coarse scale and guided by a weighting scheme, our model firstly learns to match down-sampled nodes whose vicinity points share more overlap, which significantly shrinks the search space of a consecutive stage. On a finer scale, node proposals are consecutively expanded to patches that consist of groups of points together with associated descriptors. Point correspondences are then refined from the overlap areas of corresponding patches, by a density-adaptive matching module capable to deal with varying point density. Extensive evaluation of CoFiNet on both indoor and outdoor standard benchmarks shows our superiority over existing methods. Especially on 3DLoMatch where point clouds share less overlap, CoFiNet significantly outperforms state-of-the-art approaches by at least 5% on Registration Recall, with at most two-third of their parameters. [Code]

No abstract

As the technology moves towards more human-like bionic limbs, it is necessary to develop a feedback system that provides active touch feedback to a user of a prosthetic hand. Most of the contemporary sensory substitution methods comprise simple position and force sensors combined with few discrete stimulation units, and hence they are characterized with a limited amount of information that can be transmitted by the feedback. The present study describes a novel system for tactile feedback integrating advanced multipoint sensing (electronic skin) and stimulation (matrix electrodes). The system comprises a flexible sensing array (16 sensors) integrated on the index finger of a Michelangelo prosthetic hand mockup, embedded interface electronics and multichannel stimulator connected to a flexible matrix electrode (24 pads). The developed system conveys contact information (binary detections) to the user. To demonstrate the feasibility, the system was tested in six able-bodied subjects who were asked to recognize static patterns (contact position) with two different spatial resolutions and dynamic movement patterns (i.e., sliding along and/or across the finger) presented on the electronic skin. The experiments demonstrated that the system successfully translated the mechanical interaction into electrotactile profiles, which the subjects could recognize with good performance. The success rates (mean ± standard deviation) for the static patterns were 91 ± 4% and 58 ± 10% for low and high spatial resolution, respectively, while the success rate for sliding touch was 94 ± 3%. These results demonstrate that the developed system is an important step towards a new generation of tactile feedback interfaces that can provide high-bandwidth interfacing between the user and his/her bionic limb. Such systems would allow mimicking spatially distributed natural feedback, thereby facilitating the control and embodiment of the artificial device into the user body scheme.

Establishing correspondences from image to 3D has been a key task of 6DoF object pose estimation for a long time. To predict pose more accurately, deeply learned dense maps replaced sparse templates. Dense methods also improved pose estimation in the presence of occlusion. More recently researchers have shown improvements by learning object fragments as segmentation. In this work, we present a discrete descriptor, which can represent the object surface densely. By incorporating a hierarchical binary grouping, we can encode the object surface very efficiently. Moreover, we propose a coarse to fine training strategy, which enables fine-grained correspondence prediction. Finally, by matching predicted codes with object surface and using a PnP solver, we estimate the 6DoF pose. Results on the public LM-O and YCB-V datasets show major improvement over the state of the art w.r.t. ADD(-S) metric, even surpassing RGB-D based methods in some cases.