International Conference «Mathematical and Information Technologies, MIT-2016»

28 August – 5 September 2016

Vrnjacka Banja, Serbia – Budva, Montenegro

Kamaev A.N.   Karmanov D.A.   Sukhenko V.A.   Smagin S.I.  

Using of Synthetic Data for Testing and Debugging AUV Technical Vision Systems

Reporter: Smagin S.I.

   Complexity of underwater navigation and inability to organize communication with high capacity data exchange between AUV (autonomous underwater vehicle) and operator leads to necessity of onboard technical vision systems development for improving underwater vehicle autonomy. High reliability and quality operation in various conditions is required from such systems. Their development and further usage is impossible without precise debugging and testing. Currently testing process organization and testing data obtaining becomes a serious problem.
   At present time, testing areas with markers and targets located on seabed and specifically equipped pools are used for tasks of testing and debugging. Such methods are well suited for testing of AUV’s devices and equipment, but they are not capable to provide enough amount and variety of data necessary for technical vision system testing and debugging. Besides, time required for data obtaining is too long, and that is unacceptable especially at the early stages of development.
   Virtual water areas, built using computer graphic procedural generation methods, are suggested for solving a problem of AUV’s technical vision systems testing and debugging. Proposed system for testing consists of three main parts: virtual water area generator, interface between developing system and virtual AUV and AUV simulator:
1. Virtual water area generator is the most important subsystem. Its task is building and shading of the synthetic seabed model. The process of generation is controlled by the operator through defining a small set of parameters determining seabed appearance. The seabed is built by combining three levels of detail: global low-resolution digital elevation map (built procedurally based on parameters), 3D mesh (built based on digital elevation map and 3D «smart noise» with the help of marching cubes technique) and high-frequency fractal noise (generated and added during rendering). Testing area is shaded procedurally based on surface roughness and average slope. During generation process the variety of surfaces types and their similarity to real world ones and the absence of repeating texture patterns is provided.
2. Interface between technical vision system and virtual AUV is provided by passing controlling messages (requests on photos from AUV’s cameras and control signals) from vision system and by receiving informational messages (photos with debug information and information about real AUV location and orientation) from virtual AUV.
3. AUV simulator provides the simplest physical model of AUV movement and gives possibility of testing area rendering from onboard AUV’s cameras points of view

   Using of the synthetic testing areas allows us to obtain a lot of advantages in compare with real testing areas:
1. the high speed and low cost of testing data acquisition,
2. the ability to obtain testing data from different surfaces types, starting from sand valleys and ending with rocky canyons,
3. the ability to evaluate technical vision accuracy since the investigated seabed surface is precisely known,
4. the ability to interrupt system working exactly at the moment when error  occurs,
5. the testing result repeatability.

   It is evident that using of synthetic tests does not allow us to completely abandon the real experiments, but it significantly reduces their amount. As a result, the time required for AUV’s technical vision system development is decreasing and reliability is increasing.


To reports list

© 1996-2019, Institute of computational technologies of SB RAS, Novosibirsk