Skip to Main Content
Autonomous mobile robots perform their activities in unstructured, dynamic and unpredictable environments. The robot state, the objectives to be fulfilled and the environmental conditions have influence in the variability of the system load. In the face of such environmental uncertainty, the computational requirements of recognition tasks are variable and dependent on the number of objects perceived in the scenes. It also arises that to ensure robot safety, the temporal requirements of reactive tasks have to be proportionally adjusted to the actual robot speed. Traditional robot control architectures have been designed ignoring these aspects and hence lead to extremely expensive and underutilised system designs. To overcome such drawbacks a feedback control scheduler (FCS) together with a task model that permit the adaptation of the temporal requirements of the tasks depending on the robot speed and on the environmental conditions are proposed. Furthermore, to undertake robot QoS improvement, a flexible server (FS) is integrated with the FCS in a global real-time architecture.