Learning-based haptic perception in robotic grippersdesign and applications

Resumen

The world of robotics has been growing exponentially over the last decades. Recent advances in control techniques, sensors, and actuators, as well as improvements in the ability to interpret the surrounding environment through the use of Artificial Intelligence (AI) are enhancing the concept of intelligent robots day by day. In a humanity-centered world, advances in robotics are mainly focused on the interaction between humans and robots. Based on this premise, robots should be able to generate interaction strategies for cooperation and coaction with humans, and plan motions that take into account our needs and safety. However, to perform these procedures, robots should be capable of perceiving the world around them. One of the most important senses we humans have to interact with the surrounding environment is the sense of touch. The word haptic refers to everything that is related to the sense of touch. We can define the haptic term as the sensations that encompass tactile and kinesthetic perception. Haptic perception is almost indispensable for physically interacting with the world around us. Therefore, it is critical to design end-effectors for robots with sensors that perceive sensations analogous to haptic perception. It is also necessary to interpret the perceived information to act accordingly. This thesis encompasses the design of grippers with haptic perception capabilities, and the intelligent use of the information perceived. The grippers are able to perform tasks where there is direct robot-initiated physical interaction. These tasks are numerous, e.g., grasping human limbs to carry out rehabilitation exercises, performing palpation procedures to grasped bodies, or placing sensors in specific areas. In particular, this dissertation presents three contributions related to the general basis of the thesis. Intelligent grippers are developed and used for diverse purposes: I) Fusing haptic information perceived with grippers; II) Sensing interaction forces while grasping in physical Human-Robot Interaction (pHRI); III) Placing sensorized grippers to victims with a mobile manipulator in Search and Rescue (SAR) scenarios. The present thesis is a compendium of previously published scientific articles, which form the core of the PhD. Four papers published in prestigious indexed scientific journals make up the main body of this study. The contributions addressed in this study are discussed in chapters 2, 3, and 4.