MULTIMODAL INTERACTION FOR MOBILE ROBOT GUIDANCE

摘要

Human-robot interaction is a very important issue for autonomous robots, in particular when the targeted environment is more general than the strictly constrained, fully-controlled factory automation. Modern robots are basically computers equipped with complex actuators and sensing devices: thus their native communication paradigm can be described in terms of tokens and numeric data and performed through a keyboard and a monitor or similar devices. We introduce multimodal human-computer interface and robot guidance platform allowing a human user to communicate with a robot, associate visual tags and spoken tokens to objects and ask the robot to perform actions. To illustrate our work, we implemented the presented system on a mobile robot platform, making it autonomous and able to perform navigational and object recognition and manipulation tasks. Experimental results are shown and discussed.