Indonesian Journal of Electrical Engineering and Computer Science Vol. 25, No. 3, March 2022, pp. 1344~1355 ISSN: 2502-4752, DOI: 10.11591/ijeecs.v25.i3.pp1344-1355  1344 Journal homepage: http://ijeecs.iaescore.com New lambda tuning approach of single input fuzzy logic using gradient descent algorithm and particle swarm optimization Fauzal Naim Zohedi, Mohd Shahrieel Mohd Aras, Hyreil Anuar Kasdirin, Nurdiana Binti Nordin Underwater Technology Research Group (UTeRG), Center for Robotics and Industrial Automation (CERIA), Fakulti Kejuruteraan Elektrik, Universiti Teknikal Malaysia Melaka, Melaka, Malaysia Article Info ABSTRACT Article history: Received Oct 29, 2021 Revised Jan 11, 2022 Accepted Jan 17, 2022 Underwater remotely operated vehicle (ROV) is important in underwater industries as well as for safety purposes. It can dive deeper than humans and can replace humans in a hazardous underwater environment. ROV depth control is difficult due to the hydrodynamic of the ROV itself and the underwater environment. Overshoot in the depth control may cause damage to the ROV and its investigated location. This paper presenting a new tuning approach of single input fuzzy logic controller (SIFLC) with gradient descent algorithm (GDA) and particle swarm optimization (PSO) implementation for ROV depth control. The ROV was modeled using system identification to simulate the depth system. Proportional integral derivative (PID) controller was applied to the model as a basic controller. SIFLC was then implemented in three tuning approaches which are heuristic, GDA, and PSO. The output transient was simulated using Matlab/Simulink and the percent overshoot (OS), time rise (Tr), and settling time (Ts) of the systems without and with controllers were compared and analyzed. The result shows that SIFLC GDA output has the best transient result at 0.1021% (OS), 0.7992 s (Tr), and 0.9790 s (Ts). Keywords: Gradient descent algorithm Particle swarm optimization Proportional integral derivative controller Remotely operated vehicle Single input fuzzy logic controller This is an open access article under the CC BY-SA license. Corresponding Author: Fauzal Naim Zohedi Underwater Technology Research Group (UTeRG), Center for Robotics and Industrial Automation (CERIA), Fakulti Kejuruteraan Elektrik, Universiti Teknikal Malaysia Melaka 76100 Durian Tunggal, Melaka, Malaysia Email: fauzal@utem.edu.my 1. INTRODUCTION In the underwater engineering field, remotely operated vehicle (ROV) plays an important role in underwater observation, investigation, and inspection [1]–[3]. Especially in the oil and gas industry, ROV is used to do underwater pipe inspections as well as repairing jobs. ROV normally suffered from problems that include pose recovery or station keeping, underactuated conditions, coupling issues, and communication techniques [4]. This research paper was focusing on ROV depth control or station keeping. Station keeping at a certain depth is very important for underwater exploration and inspection [5]–[7]. However, controlling ROV is difficult because of the unexpected and unpredictable underwater environment [4], [8]. This is due to the nonlinear hydrodynamics effect, coupled characters of plant equations, lack of precise models of underwater vehicle hydrodynamics and uncertainty parameters [9], [10], as well as the presence of environmental disturbances [1], [11]–[14]. Controller design, based on simple models of underwater vehicle mass and drag, generally yields unacceptable performances [15]. Linear (conventional) controller is unable to adequately control the unmanned underwater vehicle (UUV) satisfactorily [16]. Even for a one-axis motion such as vertical motion or heave motion, consistent performance for a desirable range is required. Overshoot in the system is