Argumentation-Based Incremental Learning for Home Robotics

The environment around general purpose service robots has a dynamic nature. Accordingly, even the robot's programmer cannot predict all the possible external failures which the robot may confront. This research concentrates on proposing an autonomous approach for on-line incremental learning from observation of failure states. Existing research on handling external failures typically offers special-purpose solutions depending on what has been foreseen at the design time. Furthermore, the current incremental online learning algorithms are not capable of extracting an appropriate set of hypotheses with just a few observations. These extracted set of hypotheses can be then used for classification. The proposed argumentation-based on-line incremental learning approach uses the abstract and bipolar argumentation framework to extract the most relevant hypotheses and model the defeasibility relation between them. This leads to a novel on-line incremental learning approach which overcomes the addressed problems. The resulting approach learns more quickly with a lower number of observations and also has higher final precision than other methods.