Measuring Software Complexity to Target Risky Modules in Autonomous Vehicle Systems

摘要

We established that there are metrics and related threshold values for software quality attributes like reliability, maintainability, and ease of integration. There are tools that quickly compute these values without running the application. Scatter graphs help us visualize the modules belonging to the quadrant of unreliable and hard to maintain domain. To prioritize which modules to select for analysis, the tool's standard threshold values were increased. The rationale was one of team software maturity level and the need to meet delivery of the application in four months. This helped focus the developers on the most at-risk modules. Moreover, software developers can tactically select which modules need peer-reviews based on the module's code complexity. Program managers can set realistic times to fix code based on the use of unstructured code.