Assistive feeding robot for upper limb impairment—Testing and validation

Abstract

A personal care robotic system has been developed that can provide feeding assistance to those suffering from upper limb impairment. The system introduces a novel approach for feeding that prioritizes two ideas: generalized functionality to encompass multiple feeding tasks and seamless user interaction. Additionally, the system leveraged novel computer vision ideas to incorporate functionality that was not reported in the literature. For the functional prototype, the system was comprised of an off-the-shelf six degrees of freedom robotic manipulator, two depth cameras, and an electric gripper. Furthermore, various tools used during the operation were designed and constructed using a 3D printer. The system’s software has three main operation phases: food identification, acquisition, and delivery. One of the novel features of this system is that instead of attempting to identify the food, the robot identifies the method required for acquiring the food. During testing and validation, it was found that the system had minimal identification errors, high success rates for acquisition and delivery, and a fast safety response time.