Larasmoyo Nugroho
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Larasmoyo Nugroho has a total of 16 co-authors. Here are the top co-authors by number of shared publications:
Rika Andiarti
Rocket Technology Center, National Research and Innovation Agency, Bogor, Indonesia
3 shared publicationsRini Akmeliawati
School of Mechanical Eng., University of Adelaide, Adelaide, Australia
3 shared publications- Sastra Kusuma Wijaya
Department of Physics, Universitas Indonesia (UI), Depok, 16424, Indonesia
3 shared publications Abdul Khoifan
Politeknik Elektronika Negeri Surabaya, Surabaya, Indonesia
1 shared publication- Ali Turker Kutay
Department of Aerospace Engineering, Faculty of Engineering, Middle East Technical University, 06800 Ankara, Turkey
1 shared publication Ammar Abdurrauf
1 shared publication
Latest Publications
- Open Access
- 29 February 2024
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 19 (2) , e0292539
Enhancing the landing guidance of a reusable launch vehicle by improving genetic algorithm-based deep reinforcement learning using Hybrid Deterministic-Stochastic algorithm
The PbGA-DDPG algorithm, which uses a potential-based GA-optimized reward shaping function, is a versatiledeep reinforcement learning/DRLagent that can control a vehicle in a complex environment without prior knowledge. However, when compared to an established deterministic controller, it consistently falls short in terms of landing distance accuracy. To address this issue, the HYDESTOC Hybrid Deterministic-Stochastic (a combination of DDPG/deep deterministic policy gradient and PID/proportional-integral-derivative) algorithm was introduced to improve terminal distance accuracy while keeping propellant consumption low. Results from extensive cross-validated Monte Carlo simulations show that a miss distance of less than 0.02 meters, landing speed of less than 0.4 m/s, settling time of 20 seconds or fewer, and a constant crash-free performance is achievable using this method.
- 29 February 2024
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 19 (2) , e0292539
Open AccessEnhancing the landing guidance of a reusable launch vehicle by improving genetic algorithm-based deep reinforcement learning using Hybrid Deterministic-Stochastic algorithm
The PbGA-DDPG algorithm, which uses a potential-based GA-optimized reward shaping function, is a versatiledeep reinforcement learning/DRLagent that can control a vehicle in a complex environment without prior knowledge. However, when compared to an established deterministic controller, it consistently falls short in terms of landing distance accuracy. To address this issue, the HYDESTOC Hybrid Deterministic-Stochastic (a combination of DDPG/deep deterministic policy gradient and PID/proportional-integral-derivative) algorithm was introduced to improve terminal distance accuracy while keeping propellant consumption low. Results from extensive cross-validated Monte Carlo simulations show that a miss distance of less than 0.02 meters, landing speed of less than 0.4 m/s, settling time of 20 seconds or fewer, and a constant crash-free performance is achievable using this method.
- Open Access
- 1 January 2023
- Published by AIP Publishing in AIP Conference Proceedings
- Vol. 2941 (1)
Improvement of the DDPG algorithm via twin delayed DDPG (TD3) on vertical rocket landing control system
The Double Deep Policy Gradient or DDPG Reinforcement Learning algorithm has a tendency to overestimate computations, which results in learning policies that are not optimum. When DDPG agent is applied in the rocket landing control system, high frequency chattering occurs at the main engine’s control command and this gives substantial drawbacks for the liquid propulsion life longevity. The Twin-Delayed Deep Deterministic Policy Gradient (TD3) Reinforcement Learning agent is proposed to overcame this chattering by delaying the landing rocket control command, as simulated in two dimensions. This TD3 controller is shown to be able to dampen the engine command as its achievements are contrasted with the outcomes of the DDPG one.
- 1 January 2023
- Published by AIP Publishing in AIP Conference Proceedings
- Vol. 2941 (1)
Open AccessImprovement of the DDPG algorithm via twin delayed DDPG (TD3) on vertical rocket landing control system
The Double Deep Policy Gradient or DDPG Reinforcement Learning algorithm has a tendency to overestimate computations, which results in learning policies that are not optimum. When DDPG agent is applied in the rocket landing control system, high frequency chattering occurs at the main engine’s control command and this gives substantial drawbacks for the liquid propulsion life longevity. The Twin-Delayed Deep Deterministic Policy Gradient (TD3) Reinforcement Learning agent is proposed to overcame this chattering by delaying the landing rocket control command, as simulated in two dimensions. This TD3 controller is shown to be able to dampen the engine command as its achievements are contrasted with the outcomes of the DDPG one.
- 3 November 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-7
Visually Augmented Guidance System Realization for Landing Rocket Model
UAV's auto landing based on GPS data is considered to be unprecise, because GPS inaccuracy is inherently attributed to the low rate of satellite signal update. Besides of that, GPS data does not pay attention to the condition of the area and allows landing in any hazardous place. The visually augmented precision landing (VAPL) guidance system proposed in this paper guides the landing rocket model (LRM) to conduct vertical-takeoff vertical-landing (VTVL) maneuvers by fusing data from vision sensors, GPS and even ultrasonic as navigational aids. This system uses the MAVlink protocol to communicate data between the sensor board and the flight controller. The difficulty of tracking object using visual sensor is caused by many factors i.e. light intensity, color saturation, parallax, and aspect angle segmentation. This paper addresses those obstacles by refining calibration procedure and replace area segmentation method to proportional feedback method. Unwanted inverted flight response could be dealt with the improved program algorithm. The results show that the proportional feedback method approach can reduce the error in area segmentation method significantly, from 100% error to maximum value of accuracy up to 80%. In this study, the landing rocket model could land on the target, but data errors still exist due to the mistakes produced by the sensor vision when focusing on the target while discriminating it from the larger block area. The miss distance between the LRM to the target is approximately 1 meter. Therefore, the proposed VAPL could be considered more precise than a solely GPS-guided landing system that has average miss distance 3 meters.
- 3 November 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-7
Visually Augmented Guidance System Realization for Landing Rocket Model
UAV's auto landing based on GPS data is considered to be unprecise, because GPS inaccuracy is inherently attributed to the low rate of satellite signal update. Besides of that, GPS data does not pay attention to the condition of the area and allows landing in any hazardous place. The visually augmented precision landing (VAPL) guidance system proposed in this paper guides the landing rocket model (LRM) to conduct vertical-takeoff vertical-landing (VTVL) maneuvers by fusing data from vision sensors, GPS and even ultrasonic as navigational aids. This system uses the MAVlink protocol to communicate data between the sensor board and the flight controller. The difficulty of tracking object using visual sensor is caused by many factors i.e. light intensity, color saturation, parallax, and aspect angle segmentation. This paper addresses those obstacles by refining calibration procedure and replace area segmentation method to proportional feedback method. Unwanted inverted flight response could be dealt with the improved program algorithm. The results show that the proportional feedback method approach can reduce the error in area segmentation method significantly, from 100% error to maximum value of accuracy up to 80%. In this study, the landing rocket model could land on the target, but data errors still exist due to the mistakes produced by the sensor vision when focusing on the target while discriminating it from the larger block area. The miss distance between the LRM to the target is approximately 1 meter. Therefore, the proposed VAPL could be considered more precise than a solely GPS-guided landing system that has average miss distance 3 meters.