This paper presents the novel observation model, called Modified Spherical Signature Descriptor(MSSD), capable of representing 2D image generated from 3D point cloud data. The Modified Spherical Signature Descriptor has a uniform mesh grid to accumulate the occupancy evidence caused by neighbor point cloud data. According to a kind of area such as wall, road, tree, car, and so on, the evidence pattern of 2D image looks so different each other. For the parameter learning of Convolutional Neural Network(CNN) layers, these 2D images were applied as the input layer. The Convolutional Neural Network, one of the deep learning methods and familiar with the image analysis, was utilized for the urban structure classification. The case study on CNN practice was introduced in detail in this paper. The simulation results shows that the classification accuracy of CNN with 2D images of the proposed MSSD was improved more than the traditional methods' one.

This research is a case study of underwater object tracking based on real-time recurrent regression networks (Re3). Re3 has the concept of generic object tracking. Because of these characteristics, it is very effective to apply this model to unclear underwater sonar images. The model also an pursues object tracking method, thus it solves the problem of calculating load that may be limited when object detection models are used, unlike the tracking models. The model is also highly intuitive, so it has excellent continuity of tracking even if the object being tracked temporarily becomes partially occluded or faded. There are 4 types of the dataset using multi-beam sonar images: including (a) dummy object floated at the testbed; (b) dummy object settled at the bottom of the sea; (c) tire object settled at the bottom of the testbed; (d) multi-objects settled at the bottom of the testbed. For this study, the experiments were conducted to obtain underwater sonar images from the sea and underwater testbed, and the validity of using noisy underwater sonar images was tested to be able to track objects robustly.

In the ground environment, mobile robot research uses sensors such as GPS and optical cameras to localize surrounding landmarks and to estimate the position of the robot. However, an underwater environment restricts the use of sensors such as optical cameras and GPS. Also, unlike the ground environment, it is difficult to make a continuous observation of landmarks for location estimation. So, in underwater research, artificial markers are installed to generate a strong and lasting landmark. When artificial markers are acquired with an underwater sonar sensor, different types of noise are caused in the underwater sonar image. This noise is one of the factors that reduces object detection performance. This paper aims to improve object detection performance through distortion and rotation augmentation of training data. Object detection is detected using a Faster R-CNN.