Lessons Learned from Model Checking a NASA Robot Controller

Abstract

Abstract. This paper reports as a case study an attempt to model check the control subsystem of an operational NASA robotics system. Thirty seven properties including both safety and liveness specifications were formulated for the system. Twenty two of the thirty seven properties were successfully model checked. Several significant flaws in the original software system were identified and corrected during the model checking process. The case study presents the entire process in a semi-historical mode. The goal is to provide reusable knowledge of what worked, what did not work and why.

ModelRob: A Simulink Library for Model-Based Development of robot manipulators

“…Prop 2 Eventually the end-effector follows the desired trajectory within a certain specified error bound. Sharygina et al [29] formulated 37 properties for a NASA robot controller including various forms of the above two general properties, and successfully verified 22 of them. Their robot controller is implemented in C++.…”

Trajectory Tracking Control of Robot Manipulators

“…Robot manipulators [1] [6] find many applications in many systems, for example spacecraft [2] industrial automation and life critical system. Simulink [8] is a high level model design tool is useful in many industrial application areas. Simulink [3] provides a wide range of library blocks for example arithmetic block, signal block, to name a few.…”

A Systematic Testing Approach for Autonomous Mobile Robots Using Domain-Specific Languages

“…There have been efforts to guarantee certain properties of systems using formal verification techniques [1,2]. However, the increasing number of states for complex control systems limits this approach to relatively small subsystems.…”