• contact forces
• IMU
• interaction
• soft robotics

Human intervention is essential in many dicult situations, e.g. in emergency, survey operations
and interventions in earthquake areas. In such circumstances where human life is
endangered, walking robots can come in to play, as they are able to navigate in even complex
environments. In these cases, the success of the mission depends on the control and stability
a walking robot is able to achieve, which in turn depends by the choice of rigid or adaptive
feet. After showing the state of the art of some types of walking robots and feet used, the
thesis analyses a walking foot as a grasping problem. I propose two models that reconstruct
the contact forces distribution, knowing only the shape of the foot during the contact. The
rst one is a Model-Based Method which is able to reconstruct forces from only conguration
data knowing the structure of the end-eector. The second one is a Black Box Method, where
the forces are estimated via a Neural Network.