Design and Control of an Aquatic Robot with RGB-D Sensor

Bio-inspired robotics is a relatively new field of engineering, which aims at mimicking living organisms for various applications. It focuses on building autonomous machines that can be operated in remote, inaccessible or dangerous environments, where the presence of humans or other systems is not possible. The main goal of this work is to develop a reactive control system architecture for an aquatic robot based on the feedback acquired from a Red Green Blue-Depth (RGB-D) camera to navigate and plan the path efficiently. The proposed solution is based on image segmentation and distance estimation using the depth data. This data is then imputed to a small neural network that implements a low-level reflex controller that governs the robot's body curvature in order to alter it's trajectory. The so controlled robot exhibits different behaviors like avoid obstacles in an aquatic environment and detect energy sources. The design of the aquatic robot is introduced based on segments connected with two degrees of freedom joints moved by a tendons actuation system.