Language for learning complex human-object interactions

In this paper we use a Hierarchical Hidden Markov Model (HHMM) to represent and learn complex activities/task performed by humans/robots in everyday life. Action primitives are used as a grammar to represent complex human behaviour and learn the interactions and behaviour of human/robots with different objects. The main contribution is the use of a probabilistic model capable of representing behaviours at multiple levels of abstraction to support the proposed hypothesis. The hierarchical nature of the model allows decomposition of the complex task into simple action primitives. The framework is evaluated with data collected for tasks of everyday importance performed by a human user.

Authors

• Mitesh Patel
• Carl Henrik Ek
• Nikolaos Kyriazis
• Antonis A. Argyros
• Jaime Valls Miró
• Danica Kragic

Keywords

• Hidden Markov models
• Robots
• Joints
• Probabilistic logic
• Accuracy
• Abstracts
• Data models
• Human Behavior
• Probabilistic Model
• Complex Behavior
• Hidden Markov Model
• Characteristics Of Data
• Expectation Maximization
• Transition Probabilities
• Noisy Data
• Levels Of Hierarchy
• Popular Technique
• Gaussian Mixture Model
• Object Motion
• Tracking Algorithm
• Sequential Task
• Human Motion
• Unsupervised Manner
• Cartesian Space
• Hidden Variables
• Hand Motion
• Conditional Probability Distribution
• Dynamic Bayesian Network
• Abstract States
• User Tasks
• Natural Language Descriptions
• Inference Accuracy
• Termination Condition
• Top Level
• Motion In Space
• Time Step
• Object Classification