Door recognition and deep learning algorithm for visual based robot navigation
Wei Chen, Ting Qu, Yimin Zhou et al. · IEEE International Conference on Robotics and Biomimetics (ROBIO) · 2014
In this paper, a new method based on deep learning for robotics autonomous navigation is presented. Different from the most traditional methods based on fixed models, a convolutional neural network (CNN) modelling technique in Deep learning is selected to extract the feature inspired by the working pattern of the biological brain. This neural network model has muti-layer features where the ambient scenes can be recognized and useful information such as the location of door can be identified. The extracted information can be used for robot navigation, so does the robot can approach the target accurately. In the field experiments, detecting doors and predicting the door poses such tasks are designed in the indoor environment to verify the proposed method. The experimental results demonstrate that the doors can be identified with good performance and the deep learning model is suitable for robot navigation.
Cited by 53 works. doi:10.1109/robio.2014.7090595
Authors
References
- ImageNet classification with deep convolutional neural networks
- Backpropagation Applied to Handwritten Zip Code Recognition
- Dropout: a simple way to prevent neural networks from overfitting
- What is the best multi-stage architecture for object recognition?
- Why is Real-World Visual Object Recognition Hard?
- Unsupervised Learning of Invariant Feature Hierarchies with Applications to Object Recognition
- Building high-level features using large scale unsupervised learning
- Nonlinear image representation using divisive normalization
- Learning invariant features through topographic filter maps
- A Theoretical Analysis of Feature Pooling in Visual Recognition