Performance Evaluation Framework for Arbitrary Nodes of Autonomous-driving Systems

Autonomous-driving software is increasingly based on Robot Operating System (ROS) 2, middleware for developing robotic applications. This paper proposes a performance evaluation framework for arbitrary nodes of Autonomous-driving software. The proposed framework empowers ROS 2 application developers to evaluate particular nodes requiring evaluation efficiently. The proposed framework can activate each node by considering the logic when the application is run as a whole. In addition, input data specific to each node are required for node evaluation. Because each node exchanges various types of data, generating input data to evaluate a particular node can be difficult. Thus, the proposed framework generates input data suitable for node evaluation. The proposed framework also provides a graphical user interface (GUI). A developer can easily evaluate arbitrary nodes by manipulating the proposed framework on the GUI. Experimental results show that the proposed framework successfully activates the evaluation target node and evaluates arbitrary nodes of autonomous-driving software. In addition, each functionality of the proposed framework is demonstrated to be working properly. The results of user functionality testing of the proposed framework demonstrate its usefulness.