Multi-method modeling and simulation of a face detection robotic system

摘要

Robotic systems are currently going through changes at an unprecedented pace. Although the final goal of robotic systems will necessarily focus on real world robots, it is often useful to perform simulation prior to investigation with actual robots. Modeling and Simulation of robotic applications enable evaluation of different robotic system designs prior to implementation. The present work provides a Multi-Method Modeling and Simulation study of a human-robot environment for face and skeleton detection. The study is divided in three areas that include: i) the development of the user interface for testing the face detection algorithm and collecting the appropriate data for the simulation study; ii) the physical experimental design for data collection and analysis; and iii) the simulation of the human-robot environment. Microsoft Robotics Developer Studio, Visual Studio, Kinect Sensor, and AnyLogicTM were used for defining the robotic tasks, creating the application interface, detecting the human face, and modeling and simulating the system, respectively. Agent-based, discrete event and system dynamics simulation methods were combined for the simulation of the robotic system model. The simulation model includes some of the critical variables that were not included during physical experimentation. The two major findings of this simulation study consisted of the evaluation of the impact of those critical variables on the performance of the face detection algorithm prior to the construction of the actual robot and a simulation model that demonstrates this impact.