| Popis: |
High precision tracking of all surrounding objects is a prerequisite to the autononious/senii-autononious management of multiple mobile robot systems. This task comprises not only tracking of moving objects, but also detection of obstacles or other stationary objects. We demonstrate that Expectation-Maximization applied to multiple object tracking is a promising method being flexible enough to handle the problems typical of robot-borne tracking, such as extended objects, dense object environment, mutual occlusions. Observations and preliminary conclusions based on simulated and real sensor data are discussed. |
