The objective of the research described here is to develop efficient algorithm and software tools for multiobject recognition and retrieval. This research project addresses two major issues: The first issue is the identification of features and efficient methods for feature extraction which can completely describe an object. These features can be acquired using visual and ultra-sonic sensors. The second issue is the development of efficient algorithms for the retrieval of multi-objects based on their features.; The methods and algorithms developed in this research are verified on a Unimation PUMA 560 robot. Non contact sensors (a vision and a range sensor) are employed for feature detection. The information from both sensors will be combined for feature extraction and feature mapping (sensor fusion). The sensors and the robot have been integrated for this purpose with a Pentium 133 Mhz Personal Computer.
展开▼
