Camera control for learning nonlinear target dynamics via Bayesian nonparametric Dirichlet-process Gaussian-process (DP-GP) models
IROS(2014)
摘要
This paper presents a camera control approach for learning unknown nonlinear target dynamics by approximating information value functions using particles that represent targets' position distributions. The target dynamics are described by a non-parametric mixture model that can learn a potentially infinite number of motion patterns. Assuming that each motion pattern can be represented as a velocity field, the target behaviors can be described by a non-parametric Dirichlet process-Gaussian process (DP-GP) mixture model. The DP-GP model has been successfully applied for clustering time-invariant spatial phenomena due to its flexibility to adapt to data complexity without overfitting. A new DP-GP information value function is presented that can be used by the sensor to explore and improve the DP-GP mixture model. The optimal camera control is computed to maximize this information value function online via a computationally efficient particle-based search method. The proposed approach is demonstrated through numerical simulations and hardware experiments in the RAVEN testbed at MIT.
更多查看译文
关键词
Bayesian nonparametric Dirichlet-process Gaussian-process mixture models,pattern clustering,target behaviors,motion patterns,DP-GP mixture model,optimal camera control approach,Bayes methods,Bayesian DP-GP models,data complexity,target position distributions,online information value function maximization,mixture models,function approximation,DP-GP information value function,search problems,MIT,nonlinear target dynamics learning,image sensors,information value functions,cameras,nonparametric mixture model,Gaussian processes,time-invariant spatial phenomena clustering,position control,numerical simulations,velocity field,RAVEN testbed,image motion analysis,computationally efficient particle-based search method
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络