Metrology and Standard for Advanced Perception Systems for Intelligent Manufacturing

The project develops quantitative, reproducible test methods to evaluate the robustness, accuracy and performance characteristics of advanced perception systems for general assembly and other macro scale manufacturing operations. For advanced use on the manufacturing shop floor, perception systems need to sense three dimensional structure in a dynamic, uncontrolled environment. In particular, they must be able to accurately determine the six degree of freedom (6DOF) position and orientation of parts, people, robots and other objects in motion. This will enable new applications of automation, including continuous assembly where a robot interacts with a moving assembly line. Recent and ongoing innovations in sensor technology, computer speed and perception algorithms promise to bring perception technology to a mature state, but standards and metrology are required to assist manufacturers in defining and evaluating the achievement of satisfactory performance. This will have a large impact on automated manufacturing processes, leading to increased flexibility and safer working environments.

In the National Council For Advanced Manufacturing (NACFAM) study and the Computing Community Consortium’s Manufacturing and Automation Workshop, sensors were given the most attention as being a potentially disruptive technology.  They are expected to be a technology enabler in shifting manufacturing to a more flexible, adaptable and safe state in dynamic and less structured environments. Automated manufacturing is starting to rely heavily on sensors as it moves into less structured environments and deals with a greater variety of tasks and components. Sensors are needed to locate parts, identify them and determine their properties, and manipulate them robustly in three-dimensional space. Currently, very few applications involve any sensing at all, and those that do usually involve simple measurements under highly restricted conditions. For example, it may be necessary to stop moving in order to make a measurement, or a part might have to be presented in a particular known orientation in order to be inspected. These limitations incur costs in terms of efficiency, restricted work cell layout, and the need for extra fixturing and handling equipment. This is one of the major factors that have prevented the adoption of automated manufacturing by small and medium sized businesses.

Barriers to the introduction of sensors have typically included the high costs of developing custom applications, the lack of guarantees that the sensor systems will work consistently enough for a production environment, and the fact that, in the automation community, sophisticated sensing is still in its infancy and users have little experience in using it in their applications. The project addresses these barriers by defining performance measures and metrics for sensor systems, and promoting standards that help vendors describe their products in a uniform way and help users select the right products for their applications. The project also addresses another current barrier to more widespread use of advanced automation: the need to separate robots and automation equipment from people working in the same work area. Sensors will have a major part in enabling people and robots to work together safely. The project addresses the performance metrics and standards that are needed to ensure that sensors provide the necessary levels of safety and reliability to enable human-robot collaboration.