R&D

In order to give Nymo its abilities and make it an effective and reliable autonomous working platform, we are focusing to following key technologies.

Millimeter Wave Radar

In order to perceive the surroundings, our autonomous robotic ship NYMO uses mm wave radar. To process reflected signals we are working different grouping methods like Mini Batch K-Means, K-Means, DBSCAN, OPTICS and a 13 custom boxing methods. We have worked out our own unique software that captures data from the radar, saves it, groups data points and displays them on 2D scatter plot for a better visualization. Different types of scenarios were chosen to test the radar and our software. The data stream from the radar was collected and stored in a file. Later the data file was used to tune the data analysis algorithms.

Target Tracking by Neural Predictive Control

A modeling and simulation technique for an autonomous surface vehicle (ASV) together with predictive neural network control in Simulink/MATLAB environment was analyzed. The proposed methodology for the theoretical model task is planned to use for improvement of Nymo.

The simulations confirmed high performance (smoothness, fast response and accurate control) of the predictive neural network based control. The introduced ability to predict vessel behavior in the case of strongly nonlinear model proves to be a promising solution control of ASV. From the applications viewpoint the proposed methodology offers a basis for development a digital twin of Nymo type dedicated for sea pollution detection, water environment monitoring and small cargo transportation between islands.

Block diagram for modeling of ASV control with Neural Network Predictive Controller
Input control signal (yaw moment) in case of target position (40,70). Note that neural network based artificial intelligence is capable to assure maximum signal control in the beginning of course direction stabilization and later transition to the accurate small-signal control.
Output value (heading angle) in case of target position (40,70)

ASV trajectory in case of desired target position (40,70)

Link for Article: Target Tracking by Neural Predictive Control of Autonomous Surface Vessel for Environment Monitoring and Cargo Transportation Applications

Cloud Base Operating and Digital Twin platform for autonomous ship

As digitalization is becoming an integral part of the shipping industry, more and more operators can benefit from its use. An interesting new technology that aspires to move maritime forward is the digital twin concept. Digital twins provide a virtual model of a physical ship, producing valuable insights to data.

Digital twin can help to predict events taking place at sea e.g. if we want the robot ship to travel from Tallinn to Helsinki then we would first send the digital twin virtually on its way. This is a simulation that will take into consideration various parameters wind speed, vessel capacity, wave height, sea traffic etc. As a result, the digital twin will determine the needed energy to reach the destination and helps to plan the route. Furthermore, the digital twin could also optimize mechanics, material use and energy consumption of the ship and apply different machine learning algorithms. In addition to simulation, the digital twin can also be used during the mission to predict events. It can assist different steering decisions. The digital twin will be available in the Internet cloud and could be used to design new optimized ships.

Cloud Base Operating and Digital Twin platform for autonomous ship