TY - JOUR
T1 - Performance evaluation of sensorimotor primitives using eigenvector learning method
AU - Sutton, Michael S.
AU - Larson, Amy
AU - Voyles, Richard
PY - 2001/1/1
Y1 - 2001/1/1
N2 - We present a method to evaluate the performance of an eigenvector learned sensorimotor primitive for mobile robots. At runtime, the learning system projects sensor data onto the eigenspace using eigenvectors determined in training. The result of the projection is a set of sensor values and actuator values. We developed an error metric based on comparing the projected values with the actual sensor values. When the system performs closely to how it was trained, the difference between projected and actual sensors is small and hence the error metric is small. The error increases as the performance degrades. This method is not task specific and can be used for any eigenvector learned primitive. Two example applications of the error metric are shown using wall following skills for a mobile robot. First, the metric is used as a transition cue for multi-primitive sequential tasks. Second, the error metric is used to create an adaptive system that chooses the best performing skill.
AB - We present a method to evaluate the performance of an eigenvector learned sensorimotor primitive for mobile robots. At runtime, the learning system projects sensor data onto the eigenspace using eigenvectors determined in training. The result of the projection is a set of sensor values and actuator values. We developed an error metric based on comparing the projected values with the actual sensor values. When the system performs closely to how it was trained, the difference between projected and actual sensors is small and hence the error metric is small. The error increases as the performance degrades. This method is not task specific and can be used for any eigenvector learned primitive. Two example applications of the error metric are shown using wall following skills for a mobile robot. First, the metric is used as a transition cue for multi-primitive sequential tasks. Second, the error metric is used to create an adaptive system that chooses the best performing skill.
UR - http://www.scopus.com/inward/record.url?scp=0035558819&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=0035558819&partnerID=8YFLogxK
U2 - 10.1109/IROS.2001.976293
DO - 10.1109/IROS.2001.976293
M3 - Article
AN - SCOPUS:0035558819
SN - 2153-0858
VL - 2
SP - 963
EP - 967
JO - IEEE International Conference on Intelligent Robots and Systems
JF - IEEE International Conference on Intelligent Robots and Systems
ER -