In past years NIST has sometimes served as a gateway to technological progress. Standards and metrics can be a gateway when they provide a lens by which the world can be measured and understood. The MNIST database for example has not only provided a means for benchmarking, but also for people around the world to learn about artificial neural networks through tutorials that involved using the free and easily accessible database. Robotics as a field of technology is in great need of a gateway. The cost of robotics hardware makes it challenging for researchers to build on existing research, and for educators to have the tools needed to teach robotics. From an industry perspective, our country’s difficulties in implementing automation are greatly affected by our difficulties with educating tomorrow’s roboticist.
The first thing we’d like to explore is the cheapest possible robotic system capable of meaningful robot training and/or manufacturing production (advanced assembly, pick and place). Academic journals and social media often focus on the cutting edge of technology, those capabilities that have just recently become possible, what we call the top edge of technology. While this work and the reporting of it is a necessity for the advancement of technology, the absolute capabilities of the cheapest possible technologies is what drives society forward. We call this the bottom edge of technology. As part of the preliminary work, we would like to explore the bottom edge of manufacturing robotics technology and ascertain what it is.
We are interested in exploring the creation of a testbed that is:
It is called the Modular Open-source Robotics Testbed (MORT). It will have use for robotics research in manufacturing but not exclusively. This testbed would provide a standardized setup for:
It would work in a similar way to how machine learning (ML) algorithms are tested and compared. Robotics algorithms would have a common platform for development and comparison. Benchmarks could then be created around/on the testbed and progress easily tracked through time. This could include everything from ML software algorithms to grasping and manipulation hardware where only a single piece of the testbed is substituted.
The testbed is to be built from low-cost, off-the-shelf components that are modular and not reliant on any specific hardware company. The testbed's open-source and low-cost nature means that it will not be reliant on a single manufacturer and will be accessible to a wider range of students and researchers. We hope to make robotics research, training, and education more accessible to all educational institutions.
We are currently in the process of gathering information. Our team is attempting to speak with as many researchers and institutions as possible that are involved in robotics in order to better understand the challenges that robot educators and researchers are facing. We are also trying to get a sense of what MORT should look like in order to benefit the largest community.