Runtime Verification on Robotics Systems

Runtime verification is a technique for generating monitors from formal specification of expected behaviors for the underlying system. It can be applied to automatically evaluate system execution, either on-line or off-line, analyzing extracted execution traces; or it can be used online during operation, potentially steering the application back to a safety region if a property is violated. As a so-called light-weighted formal method, runtime verification bridges the gap between system design and implementation and shorten the distance of software quality assurance between the software testing and model checking and theorem proving. Runtime verification is considered as a highly scalable and automatic technique. Most of current runtime verification research are endeavored on the program context, in other words, on the program side and falls in the implementation level. These applications limited the benefits of runtime verification that bridges the gap among types of applications. With the proliferation of embedded systems and mobile device, dynamically verifying the firmware and mobile apps becomes a new emerging area. Due to the characteristics of runtime verification technique and limitations of the robotics systems, so far, very few research and project are located in the runtime verification on the firmware of embedded systems, which appear in most of robotics systems. Robotics systems are programmed on the firmware and only observed on device. In this paper, the authors first discussed the current runtime verifications on the embedded systems with limitations. After that, a layered runtime verification framework will be presented for the firmware verification. The case study is applied on the commonly recognized educational toolkit – LEGO Mindstorm robotics systems.