Using Hill Climb Modular Assembler Encoding and Differential Evolution to evolve modular neuro-controllers of an autonomous underwater vehicle acting as a Magnetic Anomaly Detector
“…The AUV carried out a mission by following the lawnmower trajectory a short distance from the bottom. Since the registration of the AUV behavior took place in simulation conditions, the task of LMGP was to recreate the mathematical model of the AUV used during simulations 50 .…”
Section: Methodsmentioning
confidence: 99%
“…The models designed during the research reported in the current paper were prepared using a small set of input-output training data recorded during the simulations, reported in 50 . This means that the task of LMGP, as well as rival methods, was to recreate another model specified in 51 .…”
In the paper, a new evolutionary technique called Linear Matrix Genetic Programming (LMGP) is proposed. It is a matrix extension of Linear Genetic Programming and its application is data-driven black-box control-oriented modeling in conditions of limited access to training data. In LMGP, the model is in the form of an evolutionarily-shaped program which is a sequence of matrix operations. Since the program has a hidden state, running it for a sequence of input data has a similar effect to using well-known recurrent neural networks such as Long Short-Term Memory (LSTM) or Gated Recurrent Unit (GRU). To verify the effectiveness of the LMGP, it was compared with different types of neural networks. The task of all the compared techniques was to reproduce the behavior of a nonlinear model of an underwater vehicle. The results of the comparative tests are reported in the paper and they show that the LMGP can quickly find an effective and very simple solution to the given problem. Moreover, a detailed comparison of models, generated by LMGP and LSTM/GRU, revealed that the former are up to four times more accurate than the latter in reproducing vehicle behavior.
“…The AUV carried out a mission by following the lawnmower trajectory a short distance from the bottom. Since the registration of the AUV behavior took place in simulation conditions, the task of LMGP was to recreate the mathematical model of the AUV used during simulations 50 .…”
Section: Methodsmentioning
confidence: 99%
“…The models designed during the research reported in the current paper were prepared using a small set of input-output training data recorded during the simulations, reported in 50 . This means that the task of LMGP, as well as rival methods, was to recreate another model specified in 51 .…”
In the paper, a new evolutionary technique called Linear Matrix Genetic Programming (LMGP) is proposed. It is a matrix extension of Linear Genetic Programming and its application is data-driven black-box control-oriented modeling in conditions of limited access to training data. In LMGP, the model is in the form of an evolutionarily-shaped program which is a sequence of matrix operations. Since the program has a hidden state, running it for a sequence of input data has a similar effect to using well-known recurrent neural networks such as Long Short-Term Memory (LSTM) or Gated Recurrent Unit (GRU). To verify the effectiveness of the LMGP, it was compared with different types of neural networks. The task of all the compared techniques was to reproduce the behavior of a nonlinear model of an underwater vehicle. The results of the comparative tests are reported in the paper and they show that the LMGP can quickly find an effective and very simple solution to the given problem. Moreover, a detailed comparison of models, generated by LMGP and LSTM/GRU, revealed that the former are up to four times more accurate than the latter in reproducing vehicle behavior.
“…Mission Oriented Operating Suit (MOOS) InterVal Programming (IvP) is a set of open source C++ modules for providing autonomy on robotic platforms, in particular, autonomous marine vehicles [1]. The software is suited for marine robotics communication, control, and simulation when the dynamic model of the vehicle is defined [20].…”
This paper discusses mathematical model implementations and simulation in various software environments for Unmanned Underwater Vehicles (UUVs), especially biomimetic ones. Gaining accurate simulation models of UUVs is challenging due to many nonlinear phenomena that need to be analysed. Further, the sensors’ accuracy and disturbances made by the natural water environment are difficult to predict during the simulation. On the other hand, an accurate simulation model is needed during new algorithm tests provided for increased vehicle autonomy. As a result, mathematical models and their implementation into different software are analysed and discussed in this paper. The model based on nonlinear differential equations is compared to an object-oriented, physical model based on Matlab Simscape Multibody.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.