Design and Implementation of a Distance-Based, Three-Dimensional Vision System

One technical challenge in the field of computer vision is to acquire information in three dimensions from an arbitrary environment. There have been successful algorithms that focus on the precision and robustness, the 3D vision, or a generalized vision system, yet none of them performs very well in all these aspects and approach the level of human vision. The research problem of this project is to design and implement a distance-based, three-dimensional vision system, which is robust in a variety of arbitrary, dynamic environments. A RaspberryPi camera module and an ultrasonic sensor were adopted for the measurement of the surrounding. The data acquired are planned to be interpreted by the computer algorithm to create a three-dimensional model. To build the prototype, multiple pieces of equipment in the school lab were utilized, such as a laser cutter, omnidirectional wheels, servos and a RaspberryPi shield. Instructions of recreating the prototype is included in the Process section of the poster. Several qualitative and quantitative objectives were listed at the start of the project, and were assessed later by analyzing the frequency and accuracy of the parts. We look forward to a completed implementation of the project, which could be applied to future landscaping, human recognition, or robot soccer.