Performance Modelling of Autonomous Mobile Robot Software Components

When the performance of robot functionalities and robot software components is evaluated, it is usually assumed to be independent from characteristics of the robot system and the environment in which they operate, such as the sensors used and the difficulty imposed by the task and the environment. This work proposes a benchmarking methodology which models the impact of the characteristics of the robot system and its environment on the performance of functionalities and their implementation as software components. Sine measuring the performance of a software component for every combination of the variables which influence the performance would be untractable, we sample a relatively small number of combinations, conduct experiments for each of them, and estimate a statistical model of the software component performance, which we call component performance model. A performance model allows us to compare different components comprehensively and to predict the performance of a robot system given the performance models of its components. We illustrate the methodology by producing performance models for different software components implementing the functionalities necessary to build autonomous navigation in indoor mobile robots: Simultaneous Localization and Mapping (SLAM), localization, local planning, and global planning.