Online Refinement of a Scene Recognition Model for Mobile Robots by Observing Human's Interaction with Environments
This work addresses a specific issue in mobile robot navigation by enabling real-time model updates through human interaction observation, though it is incremental in its approach.
The paper tackles the problem of misclassifying traversable plants as obstacles in robot navigation by proposing an online refinement method for semantic segmentation models, which outperforms ordinary weight imprinting and achieves competitive results to fine-tuning with model distillation while reducing computational costs.
This paper describes a method of online refinement of a scene recognition model for robot navigation considering traversable plants, flexible plant parts which a robot can push aside while moving. In scene recognition systems that consider traversable plants growing out to the paths, misclassification may lead the robot to getting stuck due to the traversable plants recognized as obstacles. Yet, misclassification is inevitable in any estimation methods. In this work, we propose a framework that allows for refining a semantic segmentation model on the fly during the robot's operation. We introduce a few-shot segmentation based on weight imprinting for online model refinement without fine-tuning. Training data are collected via observation of a human's interaction with the plant parts. We propose novel robust weight imprinting to mitigate the effect of noise included in the masks generated by the interaction. The proposed method was evaluated through experiments using real-world data and shown to outperform an ordinary weight imprinting and provide competitive results to fine-tuning with model distillation while requiring less computational cost.