OBSTACLE DETECTION AND OBSTACLE AVOIDANCE ON HYBRID AUTONOMOUS UNDERWATER GLIDER (HAUG)

• Main Author: Gde Jenana Putra, A.A.
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/68692
• Institution: Institut Teknologi Bandung

Hybrid Autonomous Underwater Glider (HAUG) is an underwater vehicle that is widely used for doing some underwater missions such as monitoring potential underwater resources and finding new underwater resources. HAUG has good endurance and maneuverability compared to conventional AUV (Autonomous Underwater Vehicle) and AUG(Autonomous Underwater Glider). When it is doing a mission, HAUG is usually faced with an unknown environment such as obstacles that can be a threat to the safety of HAUG. Therefore, HAUG should detect the obstacles and then do an obstacle avoidance. Multi-beam Forward Looking Sonar (FLS) is used for obstacle detection in this work. The main issues of underwater obstacle detection are noisy data received by sonar. Therefore, by designing good obstacle detection, it will overcome those issues so that can make HAUG doing obstacle avoidance. The obstacle detection used a Frost filter for noise removal. The local image histogram entropy and thresholding are used for segmentation. This research adopted BK-Fuzzy and reactive as obstacle avoidance. The RRT* path planning is also adopted in this work as obstacle avoidance. Based on the obstacle detection simulation and experiment, the size of the moving window of the filtering process used is 17x17. This moving window is a moving matrice that contains 289 pixels of sonar data. The size of the moving window used in the segmentation process is 11x11. The upper and lower thresholding value used in this research is 0.868581 and 0.613116. Based on obstacle avoidance simulation, the RRT* path planning has a higher error value between the desired path and actual path rather than BK-fuzzy and reactive. These errors are 10.6 meters dan 22.2 meters. The experiment on obstacle avoidance using BK-fuzzy and reactive used two scenarios. In the first scenario, the vehicle will be in avoidance mode when the distance between the detected obstacle and HAUG itself is lower than 4.5 meters. In the second scenario, the vehicle will be in avoidance mode when the distance between the detected obstacle and HAUG itself is lower than 2.5 meters. The errors value of desired and actual paths is 8.3 meters and 8.4 meters for each.