Design and Implementation of Dynamic Motion Policy for Multi-Sensor Omnidirectional Mobile Arm Robot System in Human-Robot Coexisted Healthcare Workplace

This paper presents the design and implementation of an Omnidirectional Mobile Arm Robot System (O-MARS) equipped with multiple sensors. The system integrates sensor information using Riemannian Motion Policies (RMP) to generate dynamic motion policies, which are continuously transmitted to the corresponding actuators of the robot. To facilitate dynamic motion policies for the robot arm and mobile robot, this paper incorporates Eye-in-Hand and Eye-to-Hand vision applications as well as self-collision avoidance. The integration of vision and LIDAR localization guides the mobile robot towards its goal position. Additionally, a virtual model of the robot's environment is constructed, and a virtual-physical integration system is devised. Within this system, the decision-making process of the RMP involves establishing relevant connections between recognized modeled environmental obstacles and the robot's digital twins. This capability enables the mobile robot to navigate around obstacles in a familiar environment. Furthermore, the inclusion of digital twins allows for the integration of LiDAR obstacle avoidance, empowering the robot to exhibit avoidance behavior when encountering unknown obstacles. The system also incorporates human skeletons and whole-body position information to autonomously analyze collision dynamics between humans and robots, ensuring their safe coexistence in shared environments. Lastly, a physical experiment is conducted in a healthcare workplace where humans and robots coexist, with a specific focus on the transportation of medical supplies. This experiment aims to demonstrate the system's effectiveness in terms of accuracy, efficiency, and safety. This paper applies the RMP approach to various domains, including sensor integration, virtual-physical integration, and the heterogeneous application of manipulators and mobile robots. The experimental results of this paper can be found at https://www.youtube.com/watch?v=pRyd0-gvutM.