Spatial Inference - Combining Learning and Constraint Solving

摘要

In our approach to spatial reasoning we use a metric description, where relations between objects are represented by parameterised homogeneous transformation matrices with nonlinear constraints on the parameters. For drawing inferences we have to multiply the matrices and to propagate the constraints. We improve a machine learning algorithm (proposed in [1] for solving these constraints. Thereafter we present the results of combining the advantages of this enhanced machine learning approach and interval arithmetics based constraint solving.