An Embeddable Object Manipulation Framework for Assistive Robotics

Recently, robots’ employment in the ambient assisted living framework is rapidly growing. Most of the employed robots fall within the category of social robots, i.e., automata able to interact verbally with the user to be assisted, supporting caregivers in patient’s need comprehension. However, most of these, although equipped with arms for social interaction, lack manipulative abilities. In this context, the paper proposes an embeddable object manipulation framework, which consists of a set of low-complexity routines to expand functionalities on social robots, and specifically on Pepper by SoftBank Robotics, permitting its employment in assistive scenarios. Implemented routines exploit Pepper’s built-in RGB cameras to (i) identify the object to be grabbed, (ii) estimate its coordinate in the three-dimensional frame; (iii) plan the arm movement sequence, and (iv) grab the object for final recognition. The routine is designed to be fully automatic (no internet connection), preserving sensitive data stored in the robot's memory. Experimental results demonstrated a grabbing accuracy of ~ 87% for different shelf heights, demonstrating the employability of improved social robotics for daily-life assistance and ambulatorial contexts.