"Multi-scale mass-deployable cooperative robots are a next-generation robotics paradigm where a large number of robots that vary in size cooperate in a hierarchical fashion to collect information in various environments. While this paradigm can exhibit the effective solution for exploration of the wide area consisting of various types of terrain, its technical maturity is still in its infant stage and many technical hurdles should be resolved to realize this paradigm." (Choo et al., 2013)
This was a part of research projects conducted under the Basic Research Laboratory Grant from Ministry of Education, Science, and Technology in South Korea. This project is two folds: (1) simulating an application of swarm robot systems; (2) designing a software framework for the swarm robot systems to reduce the complexity of developing applications while minimizing the amount of transmitted data by adopting MapReduce paradigm. The video above is a simulation of a swarm robot system application that searches for red pillars (foraging).
This project was conducted in 2008 for my bachelor's thesis in the Department of Electrical Engineering at Seoul National University (it was more like a capstone project rather than a thesis, since the focus of the project was mainly at implementing algorithms rather than analyzing the performance of algorithms using concrete measures, e.g., recall and precision). I implemented an image retrieval system prototype that takes an image as input, and outputs most similar images from the image database.
Hao Li (Ph.D. student from CS) and I conducted a big-data analysis project using the MapReduce framework (Hadoop) for the final project of INFM718G (Data-Intensive Computing with MapReduce, by Dr. Jimmy Lin). Targeting all the news images in April 2013, we tried to rank news images based on the importance and popularity level of each news image. To do that, we extracted image features using SIFT (Scale-invariant feature transform) and constructed a graph of images using LSH (Locality-sensitive Hashing) as a means to approximate the similarity of images.