Object Recognition from Local Scale-Invariant Features
ICCV '99 Proceedings of the International Conference on Computer Vision-Volume 2 - Volume 2
Decision-theoretic active sensing for autonomous agents
AAMAS '03 Proceedings of the second international joint conference on Autonomous agents and multiagent systems
Dynamic Programming
Optimal learning: computational procedures for bayes-adaptive markov decision processes
Optimal learning: computational procedures for bayes-adaptive markov decision processes
Pattern Recognition and Machine Learning (Information Science and Statistics)
Pattern Recognition and Machine Learning (Information Science and Statistics)
Cost-sensitive feature acquisition and classification
Pattern Recognition
Point-Based Value Iteration for Continuous POMDPs
The Journal of Machine Learning Research
Probabilistic planning via determinization in hindsight
AAAI'08 Proceedings of the 23rd national conference on Artificial intelligence - Volume 2
Finding approximate POMDP solutions through belief compression
Journal of Artificial Intelligence Research
Online planning algorithms for POMDPs
Journal of Artificial Intelligence Research
An autonomous six-DOF eye-in-hand system for in situ 3D object modeling
International Journal of Robotics Research
Hi-index | 0.00 |
In active perception systems for scene recognition the utility of an observation is determined by the information gain in the probability distribution over the state space. The goal is to find a sequence of actions which maximizes the system knowledge at low resource costs. Most current approaches focus either on optimizing the determination of the payoff neglecting the costs or develop sophisticated planning strategies for simple reward models. This paper presents a probabilistic framework which provides an approach for sequential decision making under model and state uncertainties in continuous and high-dimensional domains. The probabilistic planner, realized as a partially observable Markov decision process (POMDP), reasons by considering both, information theoretic quality criteria of probability distributions and control action costs. In an experimental setting an autonomous service robot uses active perception techniques for efficient object recognition in complex multi-object scenarios, facing the difficulties of object occlusion. Due to the high demand on real time applicability the probability distributions are represented by mixtures of Gaussian to allow fast, parametric computation.