Underwater Vehicles

We are addressing some of the problems associated with using acoustic communication to transmit control information to underwater vehicles. These problems stem from the time-delays present in acoustic communication links, which if not accounted for can significantly degrade the performance of underwater vehicle control, both when a vehicle is remotely-operated by a human or when multiple vehicles are collaboratively controlling each other. Furthermore, acoustic communication delays are difficult to model and predict, as they depend on a number of time-varying environmental factors and the motion of underwater vehicle. We have adopted a method to deal with these communication delays using passivity-based control that does not require a detailed model of the communication link.

Figure 1 illustrates the major features of our approach to using acoustic communication for underwater vehicle control. The critical part of this setup is the use of the scattering transform on both sides of the acoustic communication link to alleviate the problems with the communication delay. The scattering transform effectively smoothes out the signals being sent over the communication link, thereby preventing the delays from de-stabilizing the system. More precisely, the scattering transform is used to make the communication link passive. Passivity is a property of control systems that is very desirable in that connections between passive systems can be stabilized with negative feedback. What that means for this application is that as long as the other components, the human operator and the underwater vehicle controller, are passive then the overall system will converge, e.g. the underwater vehicle will accurately follow the control specified by the human operator.

 

Figure 1. Passivity-based underwater vehicle control scheme

 

The strength of this approach is that it does not require that the length of the communication delays be predicted; the scattering transform handles any length of delay and the rest of the system can be designed as if there was no delay. This approach is also very flexible to alternate choices of input or controlled vehicles. While Figure x shows an example application of a human controlling a single vehicle, our approach could also be used for a human controlling multiple vehicles simultaneously or for multiple vehicles collaboratively controlling each other without human intervention. A couple of interesting applications of this are driving multiple vehicles in formation and using multiple vehicle to autonomously cover an area. Our contribution in this work include extending previous results in passivity-based trajectory tracking of robotic vehicles with delayed communications and applying these results to underwater vehicle control. Experiments applying work to real vehicles is expected to be completed soon (see summary).