|
Study on Combination of Multiple Localizer to Improve Localization Accuracy of Outdoor Security Robots |
|
Gi-Deok Bae, Taeyoung Uhm(KIRO, Korea), Jong-Deuk Lee(Korea Institute of Robotics and Technology Convergence, Korea), Youngho Choi(KIRO, Korea) |
| We introduce three single location recognition methods based on the characteristics of each sensor.
We introduce a multi-location recognition algorithm that can obtain more accurate location recognition results using multiple single location recognition results. |
|
|
Detecting Change to Quantify Anomalies for Robust Outdoor Surveillance |
|
Muhammad Zaigham Zaheer, Marcella Astrid(University of Science and Technology, Korea), Seung-Ik Lee(Electronics and Telecommunications Research Institute, Korea) |
| Autonomous surveillance systems is becoming the need of the hour. Several researchers have proposed the idea of one-class classification, in which only normal data is used to train a one-class classification system. However, such approaches are prone to sudden change in environment, background shifts, etc. In order to eradicate such issues, we proposed to utilize change detection algorithm for anomaly detection. Such algorithms are more robust, can utilize real anomaly examples and over come the shortcomings of the one-class classification methods. |
|
|
Fighting Against Data Deficiency Problems for Surveillance |
|
Seung-Ik Lee(Electronics and Telecommunications Research Institute, Korea), Jin Ha Lee, Marcella Astrid, Muhammad Zaigham Zaheer(University of Science and Technology, Korea) |
| Anomaly detection has recently become a popular domain the field of computer vision. In this paper, we explore two
different approaches, negative learning and pseudo-anomaly generation, to improve the anomaly classification capability of autoencoders. Experiments are conducted using a real-world surveillance dataset recorded using the real CCTVs installed in a street. Experiments suggests that while negative learning is better due to the usage of real-anomaly examples, pseudo anomaly based method also provides comparable performance with an additional benefit of not utilizing real-anomaly examples. |
|
|
Anomaly Detection based on User Feedback Learning using Multi-layered Surveillance Map |
|
H.C. Shin, Jiho Chang, Kiin Na(ETRI, Korea) |
| In this study, monitoring was performed using a number of fixed monitoring agents and mobile agents, and monitoring data was converted, transmitted, and processed into a multi-layered monitoring map to solve the problems of computational amount and communication speed. In addition, a method was implemented to continuously improve judgment performance by utilizing user feedback for normal and abnormal cases determined by the system. |
|
|
Multi-modal Object Detection, Tracking, and Action Classification for Unmanned Outdoor Surveillance Robots |
|
Kyuewang Lee, Inseop Chung, Daeho Um, Jaeseok Choi, Yeji Song, Seunghyeon Seo, Nojun Kwak, Jin Young Choi(Seoul National University, Korea) |
| This paper proposes an integrated framework of multi-modal object detection, tracking, and action classification modules for unmanned outdoor surveillance robots which are equipped with multi-modal sensors. Specifically, our integrated framework utilizes RGB, thermal, and LiDAR sensors to recognize target objects' spatial location, size, motion, and action. It processes at the speed of 30~40fps, for about 3~4 objects in each scene. To validate our integrated framework, multi-modal dataset collected by unmanned surveillance robots from test-beds is used. |
|
|
Development of Cloud-based Multi-Robot Surveillance System |
|
Jung Sik Kim, Hyoung-Rock Kim, Hyoung Seok Kim, Dong Ki Noh, Seung Min Baek(LG Electronics, Korea) |
| - Security robot is becoming more common in areas such as surveillance and patrol, to detect abnormalities and prevent crimes
- We introduce a multi-modal sensor-based intelligent system for outdoor surveillance toward long term autonomy
- The result for a real-world experiment shows efficient and stable operation in various surveillance scenarios.
- The project code is available at https://github.com/lge-robot-navi. |
|