|
|
| Figure 1. Overall concept: The basic setup of the system. |
As robots are interacting with their environment more, robot vision needs to improve to real-time rates so that the robot has a chance to see its surroundings and act. Currently, passive stereo vision systems are very accurate for distances from zero to three feet. Active stereo vision is good for distances greater than fifteen feet. A real-time, accurate solution is needed for robots, especially in the three to fifteen foot range.
Objectives:The goal of this project is to develop a system using position sensitive detectors (PSDs) and FPGAs in order to build a real-time system that outputs a three-dimensional picture of the scene in front of the system. The system should be accurate within the range of zero to fifteen feet and give an output at real time or at least 30 frames per second.
Technical Approach:The basic system will be set up with two PSDs in parallel. A point laser will be placed between the two PSDs, facing the scene. The laser will scan the entire scene. At each point, the PSDs will generate x and y coordinates which will be applied to stereo equations in order to find the depth. The laser takes away the correspondence problem of passive stereo vision.
FPGAs will be used to do the necessary computations to find the x and y coordinates and also to compute the stereo vision equations.
Work is in progress.
None This research is being conducted at the