There are several challenges that must be overcome before robots can be deployed into real-world applications. Robots must be able to operate in complex, uncertain, and dynamic environments. They must be capable of providing useful services, in a robust manner, over long periods. And they must be able to interact with one or more non-technical users in a natural and productive way. For a robot to meet these criteria it must be continually aware of its surroundings, it must effectively process incoming information, and it must be capable of systematically responding to multiple commands. This thesis presents a set of visual and laser-based capabilities that provide robots with sufficient awareness and autonomy to greatly reduce the need for supervision and control by human users. By applying these capabilities to a robot-control architecture, robots can be directed to engage in a variety of realistic service-type applications with minimal effort. To validate this architecture, we show that it can allow multiple users to simultaneously request and receive services from a single robot, it can allow a single user to interact naturally with a single robot, and it can allow a single user to request and receive services from a team of robots. We demonstrate our approach on one or more Pioneer 3DX mobile robots, performing service tasks in a real-world environment.
展开▼