Task specification for predictive control and estimation in sensorbased robot applications
Robots are a cornerstone of industrial automation, and their importance is expected to further increase in the future. At the same time, the environment in which robots operate is dramatically changing: instead of executing a single task in very structured and fixed production lines, more and more companies need a very flexible deployment of robots. Instead of being locked up in their cage, with work pieces going in and out in a strictly defined way, robots are now being deployed in environments closer to human co-workers, with the position of work pieces much less accurately defined. Hence, robots need to perform multiple and more complex tasks, involving extensive sensing such as vision and force sensing.
Deploying robots for such tasks is challenging for three reasons: (i) sensing and estimation become an indispensable part of the control loop, as the robot initially has little information about the environment; (ii) the control itself becomes more challenging because both the complexity of the tasks and the required performance increase; (iii) the effort required for (re)programming the robot for its different tasks should be kept low, since the programming effort is a major cost factor and plays a large role in the economic viability of a robot application.
This project addresses the above challenges by using advanced control and estimation algorithms that are made available to the application programmers using a simple task specification language.