Geometric workcell modeling for robot control and coordination

The expansion of robotics to new industries and advances in technology brings them into closer proximity to humans which necessitates careful robot motion in relation to the environment. This report addresses the need for better geometric modeling of robotic workcells for control and coordination in terms motion planning, obstacle avoidance, and collision detection. With the realization that input modeling parameters greatly influence the type and quality of the resultant output state of the model, input/output parameters are closely examined in this report. This examination puts particular emphasis on the relationship between geometric representation and separation distance. Modeling techniques are then studied and compared to determine one which will accomplish a set of requirements for a complex robotic workcell. The Trauma Pod is a remotely controlled robotic operating room where the patient is the only human in the room. In this report, the Trauma Pod served as a case study for applying modeling techniques to fulfill requirements though a modeling framework. The requirements for the Trauma Pod included supporting collision detection and distance calculation between high resolution geometry and providing other custom features such as calculation of manipulator self-collisions. A general set of requirements were then established, and similar capabilities developed for the Trauma Pod were extended to general robotic workcells for satisfaction of those requirements. This established a set of modeling tools for a geometric workcell modeling framework. Accomplishment of all Trauma Pod modeling requirements proved the validity of the modeling technique chosen in this report. Furthermore, the chosen modeling technique has been implemented into Operational Software Components for Advanced Robotics (OSCAR) to provide powerful modeling capabilities to general robotic workcells. This work has not only proven to dramatically improve the collision detection capability of OSCAR, but it also provides potential benefit to robot motion planning and obstacle avoidance.