A novel method for learning policies from constrained motion.
Matthew HowardStefan KlankeMichael GiengerChristian GoerickSethu VijayakumarPublished in: ICRA (2009)
Keyphrases
- preprocessing
- learning scheme
- cost function
- detection method
- unsupervised learning
- active learning
- prior knowledge
- pairwise
- learning algorithm
- machine learning
- fully automatic
- motion model
- dynamic programming
- clustering method
- supervised learning
- segmentation method
- detection algorithm
- feature selection
- high accuracy
- support vector
- d objects
- support vector machine
- neural network
- image segmentation
- optical flow
- image sequences
- reinforcement learning