Download now Free registration required
It is generally agreed that learning, either supervised or unsupervised, can provide the best possible representation of a known environment. However, to operate in a dynamic environment, a mobile robot should be equipped with the ability to continuously learn novel objects and act correspondingly after that. This task is given the name novelty detection and novel objects usually refer to pattern groups that the robot has not experienced before. The framework for novelty detection generally consists of two stages. In the first stage, the robot visually explores the environment and acquires a model for it by means of some kind of learning. At the second stage, the robot uses the model to inspect the environment and identify percepts that do not fit into it.
- Format: PDF
- Size: 331.6 KB