Acoustic Image Models for Obstacle Avoidance with Forward-Looking Sonar
Long-range forward-looking sonars (FLS) have recently been deployed in autonomous unmanned vehicles (AUV). We present models for various features in acoustic images, with the goal of using this sensor for altitude maintenance, obstacle detection and obstacle avoidance. First, we model the backscatter and FLS noise as pixel-based, spatially-varying intensity distributions. Experiments show that these models predict noise with an accuracy of over 98%. Next, the presence of acoustic noise from two other sources including a modem is reliably detected with a template-based filter and a threshold learned from training data. Lastly, the ocean floor location and orientation is estimated with a gradient-descent method using a site-independent template, yielding sufficiently accurate results in 95% of the frames. Temporal information is expected to further improve the performance.
KeywordsSonar Acoustic image analysis
- 1.Folkesson, J., Leonard, J., Leederkerken, J., Williams, R.: Feature tracking for underwater navigation using Sonar. IEEE/RSJ International Conference on Intelligent Robots and Systems Proceedings (2007)Google Scholar
- 2.Horner, D., Yakimenko, O.: Recent developments for an obstacle avoidance system for a small AUV. Paper presented at IFAC Conference on Control Applications in Marine Systems, Bol island Brac, Croatia (2007)Google Scholar
- 5.Haralick, R.M., Shapiro, L.: Computer and Robot Vision, vol. 2. Addison-Wesley, Boston, MA (1993)Google Scholar