An online bridge between model-based and sensor-based path-planning under some self-positioning error

We discuss a bridge between model-based and sensor-based path-planning for a mobile robot in the presence of a self-positioning error. A model-based path-planning algorithm usually selects an arbitrary deadlock-free path in a computer world and then a mechanical vehicle faithfully follows the path in a real world. However if and only if the vehicle misunderstands its self-position by slipping and dead reckoning errors, it occasionally meets an unexpected obstruction since a selected path is to avoid all obstacles. As a result, the mechanical vehicle immediately falls into a deadlock in a real world. To avoid an unexpected obstacle in a real world, the vehicle obliges to use an online sensor-feedback function. However if the vehicle depends upon a careless sequence of online behaviors, it will join another deadlock, i.e., aperiodic sequence of online behaviors. To overcome this drawback, we propose three kinds of online bridges in order to generate the deadlock-free sequences of online behaviors along the path selected in an off-line path-planning algorithm