Medical palpation autonomous robotic system modeling and simulation in ROS/Gazebo

Manual medical palpation task performed by a human doctor is characterized by complexity of multiple source data fusion and decision subjectivity, which significantly depends on a doctor experience and sensitivity. Therefore automated robotized approach could provide more reliable and independent results relative to a manual palpation. This paper presents a model and the Gazebo simulation of an autonomous robotic system that performs palpation of an arbitrary soft surface. The system uses a modified KUKA IIWA LBR manipulator model, equipped with a spherical indenter and a new force sensor plugin. In order to perform a palpation procedure, the system gathers data from a virtual Kinect camera and computes points for palpation based on an acquired point cloud. We introduce an implementation of soft bodies features with variable stiffness to the Gazebo simulator based on DART physics engine capabilities. The developed system was tested in the Gazebo simulator by simulating palpation of abdomen soft model with variable stiffness over its surface. The system successfully performed the palpation and was able to detect an area of high stiffness.