Eating-Assistive Robotic System with Autonomous Feeding Framework and Active Compliance Control

Eating is one of the most important activities to live a healthy and independent life. Although numerous eatingassistive devices have been developed, a significant number of people with disabilities are still not able to eat independently due to the complex interaction patterns and lack of security of these devices. To solve these interaction and security problems, we developed an eating-assistive robotic system with autonomous feeding framework and active compliance control system. The autonomous feeding framework enables our robot to identify and locate food, track position and pose of the user's mouth and autonomously take and deliver food. The active compliance control system uses the robot's dynamic model and motor current to perform joint-space impedance control, thus achieving flexible contact between the robotic arm and user, ensuring user safety. Users can control the robot via voice dialogue, button presses or a simple touch on the screen, reducing the complexity of user-robot interaction methods. We performed tests with 1 participant with upper-limp disability and 4 other elderly participants. They all successfully finished a meal independently using the system and rated highly for the system's security and ease of use.