This study aims to develop a single mobile robot capable of navigating confined spaces characterized by non-circular cross-sections, embedded obstacles, and bends that are difficult for human access. To this end, we propose the Multi-Flexible Wheel Robot using brushes and elastic springs. The robot can adapt its body shape to the surrounding geometry and rotate along the meridional direction, enabling it to traverse a wide range of complex environments. This paper presents the conceptual design and driving model of the Multi-Flexible Wheel Robot, along with experimental results obtained using an initial prototype in confined space locomotion tests. In the tests, the robot traversed a straight duct with a 250 mm inner diameter and 1000 mm length in 13 s, a similar duct containing obstacles in 23 s, and a duct with varying cross-sectional geometry in 19 s. These results demonstrate the robot can transmit rotational force to multiple brushes via a deformable ring-shaped elastic structure, enabling it to autonomously navigate confined spaces that include non-circular cross-sections, obstacles, and bends.

For 5G massive MIMO systems, hybrid beam-forming (HB) has been investigated thoroughly because it is composed of analog beamforming (AB) and digital beamforming (DB), and can reduce the number of baseband and RF circuits drastically. However, HB still has a major problem of pilot contamination and thus suffers from serious degradation of the channel estimation accuracy, when the same pilot signals are shared among multiple user terminals. To cope with this problem under time-varying channel conditions, we propose a method for adaptively predicting downlink (DL) channels while mitigating the pilot contamination. Considering time division duplex (TDD) multiuser MIMO communications, the proposed scheme firstly estimates an average angle of arrival (AoA) for each user’s cluster by applying multiple signal classification (MUSIC) to the received (Rx) pilot signals, which can reconstruct the channel frequency responses of the interfering users and then extracts the desired channel frequency responses from the combined ones. Since the desired channel frequency responses can be considerably extracted, the proposed scheme coarsely compensates for time-variation of the Rx signal by predicting the Doppler effect of each cluster. The residual time-variation can be tracked and removed using detected signals of all the streams in a decision-directed manner. Finally, the estimated total time-variation is used to predict the DL channel frequency response. Computer simulations demonstrate that the proposed scheme can accurately track the time-varying channels while mitigating the pilot contamination.

We develop a deep architecture to learn to find good correspondences for wide-baseline stereo. Given a set of putative sparse matches and the camera intrinsics, we train our network in an end-to-end fashion to label the correspondences as inliers or outliers, while simultaneously using them to recover the relative pose, as encoded by the essential matrix. Our architecture is based on a multi-layer perceptron operating on pixel coordinates rather than directly on the image, and is thus simple and small. We introduce a novel normalization technique, called Context Normalization, which allows us to process each data point separately while embedding global information in it, and also makes the network invariant to the order of the correspondences. Our experiments on multiple challenging datasets demonstrate that our method is able to drastically improve the state of the art with little training data.

We present a novel deep architecture and a training strategy to learn a local feature pipeline from scratch, using collections of images without the need for human supervision. To do so we exploit depth and relative camera pose cues to create a virtual target that the network should achieve on one image, provided the outputs of the network for the other image. While this process is inherently non-differentiable, we show that we can optimize the network in a two-branch setup by confining it to one branch, while preserving differentiability in the other. We train our method on both indoor and outdoor datasets, with depth data from 3D sensors for the former, and depth estimates from an off-the-shelf Structure-from-Motion solution for the latter. Our models outperform the state of the art on sparse feature matching on both datasets, while running at 60+ fps for QVGA images.

We propose a new window search method to realize real-time object detection. Our method generates windows adaptively for objects' shapes and scales to detect various size objects. It also achieves real-time window search by using fast estimation of object's location based on existence probability of an object. Experiment results demonstrate that the proposed method reduces the number of windows drastically compared with exhaustive search. Furthermore, our method reduces the processing time while maintaining recall compared with the state-of-the art method when the numbers of searched windows are same in Pascal VOC 2007 dataset.

A wireless 3-D NoC architecture for CMPs, in which the number of processor and cache chips stacked in a package can be changed after the chip fabrication, is proposed by using the inductive coupling technology that can connect more than two known-good-dies without wire connections. Each chip has data transceivers for uplink and downlink in order to communicate with its neighboring chips in the package. These chips form a single vertical ring network so as to fully exploit the flexibility of the wireless approach that enables us to add, remove, and swap the chips in the ring. To avoid protocol and structural deadlocks in the ring network, we use the bubble flow control which is more flexible and efficient compared to the conventional VC-based deadlock avoidance. We implemented a real 3-D chip that has on-chip routers and inductive-coupling data transceivers using a 65nm process in order to show the feasibility of our proposal. The vertical bubble flow control is compared with the conventional VC-based approach and vertical bus in terms of the throughput, hardware amount, and application performance using a full system CMP simulator. The results show that the proposed vertical bubble flow network outperforms the VC-based approach by 7.9%-12.5% with a 33.5% smaller router area.