Automatic training data selection for sensorimotor primitives

Sequencing sensorimotor primitives to achieve complex behaviors can simplify programming of robotic systems. Using programming by demonstration to code the component primitives can further simplify the process. Learning methods employed in programming by demonstration require comprehensive data sets, which place a significant burden on the user during demonstration. We present a generalized method whereby training sets can be automatically filtered, freeing the user from knowledge of the underlying learning method. We achieve this by first capturing the characteristic behavior for a demonstrated task, then determining a measure of distance from that behavior. With this information, data sets can be analyzed to determine whether a particular moment of demonstration is "good" and should be included in the final training set. Results from programming by demonstration of left wall-following on a mobile platform are presented. Additionally, we present a method for on-line performance analysis that takes advantage of the characteristic behavior identified in the filtering process.

Authors

• Amy C. Larson
• Richard M. Voyles

Keywords

• Training data
• Robot programming
• Robot sensing systems
• Mobile robots
• Robotics and automation
• Learning systems
• Filtering
• Artificial neural networks
• Robustness
• Roads
• Automatic Selection
• Training Set
• Distancing Measures
• Behavioral Characteristics
• Robotic System
• Mobile Platform
• Neural Network
• Performance Measures
• Artificial Neural Network
• Actuator
• Performance Metrics
• Sensor Data
• Filtering Method
• Variety Of Situations
• Correct Behavior
• Sensor Readings
• Position Of The Robot
• Contradictory Information
• Key Sensor
• Ideal Trajectory
• Demonstration Of Behavior