Authors

Matthew Bunting, Matthew W Nice, Alex Richardson, Jonathan Sprinkle, Daniel B Work

Abstract

A method to rapidly deploy test software in a multipurpose Cyber-Physical System (CPS) platform is presented in the form of metadata called an app. Logistics in conducting an experiment involving a CPS can be challenging and is often minimally discussed in research results. Generally papers focus mainly on theory, experimental data, and analysis, but CPS research often requires careful validation and deployment efforts. A platform like a self-driving car is also likely to span various experiment use cases, so software must be cleanly uninstalled to avoid conflict in subsequent experiments. Managing a multipurpose yet safety-critical CPS between software changes can add a tremendous amount of setup time that only experts of the particular platform may be able to perform. The concept of experiment apps is presented to reduce experiment setup time in a CPS. The app structure is built to be domain-specific, targeting an open-source self driving vehicle ecosystem developed by Vanderbilt. The effort is done to further democratize the CPS development ecosystem by reducing the burden of software integration.