Collaborative Vehicles in Future Naval Missions, Obstacle Detection and Avoidance

In this Keynote paper, the authors attempt to provide some overarching view of the needs for vehicle collaboration based on future Naval missions. Collaboration between differing types of autonomous vehicle, surface, ground, aerial and underwater will be required to achieve the utility in operations promised by the concepts to date. At NPS we are also working on obstacle detection and avoidance for small AUVs which is a subject also discussed here. Recent advances in the development of low cost forward looking sonar arrays, has enabled the class of small Unmanned Underwater Vehicles to exhibit a capability for obstacle detection and avoidance. At NPS. the authors have studied the problems involved both using simulation models and through in water experimentation and validation. This paper reviews the concept of obstacle detection using a small “Blazed Array” forward looking sonar (FLS), illustrates the techniques used to analyze images obtained from an FLS, and perform threat assessment. The implementation of an avoidance controller in the NPS ARIES vehicle will be described along with a discussion of methodologies for vertical plane avoidance maneuvering. One particular strategy has been implemented and tested in the Underwater Test Range at Keyport, WA. The experiments performed will be discussed and analyzed. We show that one of the problems encountered arises when parts of the seabed are occluded from the sonar view. This leads to the notion of an uncertainty map being obtained from the FLS and used to drive the vertical response of the vehicle. Occlusion maps are built from the FLS data, and used to provide added maneuvering commands based on uncertainty. Vehicle response lags, normally a consideration with normal avoidance commands are mitigated using the FLS capability to look ahead.