Search

Object detection plays an important role in autonomous driving systems. LiDAR is widely used in autonomous driving vehicles and robots as a sensor for environmental perception. Recently, the development of computational power and deep learning technology makes it possible to classify and locate objects from LiDAR point cloud in a single end-to-end learnable network. However, objects are sparsely distributed in large point cloud field, and are always been partly scanned by LiDAR, which pose a challenge for accurate and rapid object positioning and classification from the raw point cloud. In this paper, we introduce a new single-shot refinement neural network for fast and accurate 3D object detection from the raw LiDAR point cloud. Firstly, we exploit self-attention mechanism in main object detection branch to enhance object feature representation. Secondly, we apply deformable convolution for learning adaptive receptive fields to fully capture the features of rotating and partially visible objects. Thirdly, an object refinement branch is introduced to produce a finer regression of objects upon the primary estimation from the main detection branch. All proposed modules have been proven to effectively improve the accuracy of object detection. Our method is evaluated on KITTI 3D detection benchmark and achieves state-of-the-art results while maintains real-time efficiency. Furthermore, real-time test in autonomous driving vehicle demonstrates that our method is robust to 16 channels LiDAR and can meet the demands of accuracy, efficiency, and visibility of object detection in various urban scenarios.

We propose a multielectrode circular position-sensitive device that does not have any discontinuity in the detection region and has multiple electrodes for current output. We have also manufactured a prototype device. The device has a circular detection region that consists of photodiodes radially aligned with regular pitch and a resistance line connecting the outer edges of the photodiodes. All of the photocurrent excited in the photodiodes flows into the resistance line and is extracted from the electrodes connected to the resistance line at even intervals. There is no insensitive region for current output in this device since the detection region and the resistance line are separated; thus, detection with high resolution is possible at every angular position. We conducted a simulation of the detection characteristics and will propose a computational method for the position that uses four-electrode outputs in this paper. We confirmed that it is effective in minimizing the error from both the simulation and the experiment. We will also propose a measurement system of absolute angle using this method, confirming an error of less than 0.1° in the computation results.

We developed remaining service life diagnostic technology of phenol insulators for circuit breakers. Phenol insulators were analyzed by chemical evaluation such as ion concentration and coloration, and the results were diagnosed by the Mahalanobis-Taguchi system (MTS), which is a pattern information technology. Particularly effective items for diagnosis included ions of nitric acid, sulfuric acid, chloride, and calcium. Since there was a good relationship between the results diagnosed by MTS and the actual measurement results of surface resistivity, we found that applying MTS to the chemical evaluation results allowed deterioration diagnosis of phenol insulators instead of surface resistivity, which couldn't be measured on-site. Time was defined as the end of the service life when electrical discharge was initiated and remaining service life was predicted. The linear relationship (master curve of service life) was clarified between the real number value of elapsed time and the logarithmic value of surface resistivity at 50% RH, and remaining service life was predicted from the diagnosis results by MTS and the year in which the master curve and the threshold value intersect. Surface resistivity (threshold value) for electrical discharge initiation was derived from creepage distance, permittivity of phenol insulators, frequency, etc. We substantiated that the remaining service life of the phenol insulators could be predicted by this technology because a good relationship existed between the check examination results of the electric discharge initiation in the laboratory and the prediction results.