Domain Adaptation For Mobile Robot Navigation
暂无分享,去创建一个
An important challenge in outdoor mobile robotic perception is maintaining terrain classification performance throughout the extremely variable conditions that we may wish a robot to operate under. Outdoor robots operate in a series of “environments” that consist of diverse terrain, vegetation, weather, and lighting conditions. A physical robot does not randomly jump between environments; typically it will operate for long stretches of time in one particular environment, making it advantageous to adapt the robot’s performance to its current environment.
[1] Jitendra Malik,et al. SVM-KNN: Discriminative Nearest Neighbor Classification for Visual Category Recognition , 2006, 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'06).
[2] Steffen Bickel,et al. Transfer Learning by Distribution Matching for Targeted Advertising , 2008, NIPS.
[3] H. Shimodaira,et al. Improving predictive inference under covariate shift by weighting the log-likelihood function , 2000 .
[4] James Andrew Bagnell,et al. Learning in modular systems , 2010 .