This is the chapter where we start to combine robotics and artificial intelligence to accomplish some of the tasks we have laid out so carefully in the previous chapters. The subject of this chapter is image recognition – we will be teaching the robot to recognize what is a toy, and what is not a toy, so that it can then decide what to pick up and what to leave alone. We will be using convolutional neural networks as a machine learning tool to separate objects in images, recognize them, and locate them in the camera frame, so that the robot can then locate them.
In this chapter, we will cover the following topics:
- The basics of image recognition: what is an image?
- Artificial neural networks: what is a neuron, and how do they work?
- Training neural networks using stochastic gradient descent
- Image processing...
