Model-based Object Detection

Model-based Object Detection for Navigation Researchers Tae-Bum Kwon (Ph.D. chokr@haksan.dsc.ac.kr) Period of research 2006. 10. ~ 2007. 7. Research Background Some objects such as chairs, tables, etc. can be exploited as a landmark or goal position. A robot should avoid collision with other objects. Object detection with only a range sensor is very difficult. The model information and vision sensor data can help a robot detect an object.              Research Objectives Model-based object detection using both vision and range sensors or using only a vision sensor. Obstacle avoidance through object detection. Landmark-based localization through object detection. Research Output Demo Movies Paper 1 :  Tae-Bum Kwon, Jae-Bok Song, Munsang Kim, Table Recognition through Range-based Candidate Generation and Vision-based Candidate Evaluation, Proc. of Int. Conf. on Advanced Robotics, pp.918-923, 2007.10.   Paper 2 :  Tae-Bum Kwon, Jae-Bok Song, Recognition of Objects with Legs Using Vision-based Candidate Generation and Evaluation, Proc. of Int. Conf. on Ubiquitous Robots and Ambient Intelligence, pp. 264-269, 2007.11.   * Last updated: 2012. 4. 30