최근에 개발된 시스템 온칩 프로세서는 통합 성능을 요구하는 작업의 가능성을 제공하였으며, 이러한 작업은 예전에는 우수한 성능을 가진 컴퓨터의 도움만으로 수행 할 수 있는 것이었다. 본 논 문에서는 실제 환경 하에서 자율 이동 장치의 GPS 위치측정을 개선하기 위해 임베디드 영상처리기 법을 활용하는 고급 제어 시스템을 소개한다. 메인 컨트롤 시스템은 Raspberry PI 개발 보드에 통합 된 ARM(SoC.) 아키텍처를 기반으로 한다. 제시한 제어 시스템은 실시간 비디오 캡처, 전력-효율적 이미지 처리 작업, 예를 들어 (임계 값 처리, 이진화, 모션 감지 등) 및 비디오와 같은 스트리밍 결과 이미지를 처리 할 수 있다. GPS 정밀도는 WAAS(EGNOS) 위성을 활용하여 다만 3 미터의 정밀도를 제공 할 수 있다. 제안한 솔루션은 도로와 보도의 경계를 감지하기 위해 GPS 솔루션 및 임베디드 이 미지 처리를 사용한다. 일부 도로나 통로가 길가의 흰색 선을 제공하지 않기 때문에 제시한 알고리 즘은 길가의 흰색 선을 검출하지 않고 보편적인 도로나 보도를 감지한다. 제안한 시스템은 소형 이 동장치에 사용할 수 있다. 예를 들어, 생산 공간 사이의 긴 거리를 가진 중공업 산업 단지에서 부품 수송을 위한 이동장치 등에 사용할 수 있다.

This paper presents a new benchmark system for visual odometry (VO) and monocular depth estimation (MDE). As deep learning has become a key technology in computer vision, many researchers are trying to apply deep learning to VO and MDE. Just a couple of years ago, they were independently studied in a supervised way, but now they are coupled and trained together in an unsupervised way. However, before designing fancy models and losses, we have to customize datasets to use them for training and testing. After training, the model has to be compared with the existing models, which is also a huge burden. The benchmark provides input dataset ready-to-use for VO and MDE research in ‘tfrecords’ format and output dataset that includes model checkpoints and inference results of the existing models. It also provides various tools for data formatting, training, and evaluation. In the experiments, the exsiting models were evaluated to verify their performances presented in the corresponding papers and we found that the evaluation result is inferior to the presented performances.

This paper proposes a novel method for detection of hand raising poses from images acquired from a single camera attached to a mobile robot that navigates unknown dynamic environments. Due to unconstrained illumination, a high level of variance in human appearances and unpredictable backgrounds, detecting hand raising gestures from an image acquired from a camera attached to a mobile robot is very challenging. The proposed method first detects faces to determine the region of interest (ROI), and in this ROI, we detect hands by using a HOG-based hand detector. By using the color distribution of the face region, we evaluate each candidate in the detected hand region. To deal with cases of failure in face detection, we also use a HOG-based hand raising pose detector. Unlike other hand raising pose detector systems, we evaluate our algorithm with images acquired from the camera and images obtained from the Internet that contain unknown backgrounds and unconstrained illumination. The level of variance in hand raising poses in these images is very high. Our experiment results show that the proposed method robustly detects hand raising poses in complex backgrounds and unknown lighting conditions.

In this paper, a robot vision technique is presented to detect obstacles, particularly approaching humans, in the images acquired by a mobile robot that autonomously navigates in a narrow building corridor. A single low-cost color camera is attached to the robot, and a trapezoidal area is set as a region of interest (ROI) in front of the robot in the camera image. The lower parts of a human such as feet and legs are first detected in the ROI from their appearances in real time as the distance between the robot and the human becomes smaller. Then, the human detection is confirmed by detecting his/her face within a small search region specified above the part detected in the trapezoidal ROI. To increase the credibility of detection, a final decision about human detection is made when a face is detected in two consecutive image frames. We tested the proposed method using images of various people in corridor scenes, and could get promising results. This method can be used for a vision-guided mobile robot to make a detour for avoiding collision with a human during its indoor navigation.

This paper presents a localization system using ceiling images in a large indoor environment. For a system with low cost and complexity, we propose a single camera based system that utilizes ceiling images acquired from a camera installed to point upwards. For reliable operation, we propose a method using hybrid features which include natural landmarks in a natural scene and artificial landmarks observable in an infrared ray domain. Compared with previous works utilizing only infrared based features, our method reduces the required number of artificial features as we exploit both natural and artificial features. In addition, compared with previous works using only natural scene, our method has an advantage in the convergence speed and robustness as an observation of an artificial feature provides a crucial clue for robot pose estimation. In an experiment with challenging situations in a real environment, our method was performed impressively in terms of the robustness and accuracy. To our knowledge, our method is the first ceiling vision based localization method using features from both visible and infrared rays domains. Our system can be easily utilized with a variety of service robot applications in a large indoor environment.