Real-Time Embedded Architecture for Pervasive Robots

The integration of embedded real-time control systems with the physical world via sensors and actuators (pervasive computing systems) is creating a nascent infrastructure for a technical, economic and social revolution. In most advanced real-time control applications, the tasks have different criticality, flexible timing constraints and variable execution time. For instance, autonomous mobile robots perform their activities in dynamic and unpredictable environments. The environmental conditions have influence in the variability of system load. On the one hand, the computational requirements of recognition tasks are variable and dependent on the number of objects perceived in scenes. On the other hand, the application processes are executed at different frequencies, with varying periods and deadlines that are dependent on application parameters. To tackle the computational variability of immersed robots in environment an adaptive real-time architecture is proposed in this paper.