The paper presents a joint sensing strategy that combines tactile probing and range imaging for the mapping of the elastic properties that characterize 3D deformable objects. A feedforward neural network architecture is employed in an original manner to model the complex relationship between the surface deformation and the forces exemplified in non-rigid bodies. Experimental results are presented for objects made of materials with different elastic behaviors and for their different deformation stages.
Ana-Maria Cretu, Pierre Payeur, Emil M. Petriu