Groke Pro aims at helping seafarers during times when the watchkeeping might be difficult with current systems onboard. We also try the take the benefits of each individual information source and put those all together to provide even better overall situation picture.
There are many purposes for our system, first is of course to help crew in watchkeeping and bring additional aids to detecting different obstacles under different navigation conditions.
Groke Pro system is also one of the most essential systems in future when going towards more automated navigation. No matter what the automated navigation system is it will always need our type of system to tell what is happening around the vessel.
We are also working towards so called periodical unmanned bridge approach where especially on long open sea voyages human watchkeeping could be partially replaced with machine vision. This would also be beneficial for crew when they would get additional rest time.
Of course first we have the safety benefits and the additional help our solution brings into watchkeeping. But in addition to the safety aspect, we are also looking at brining something called Groke Cloud into market in near future and that will utilise the information generated by the Groke Pro solution. So when investing into Groke Pro the client is also enabling the possibility to utilise the system in future software based add-on solutions.
The Groke Cloud solution is for the office side use and will provide valuable information from vessel to shore. If you are interested in this then we can separate session to discuss more about this opportunity as well.
We have couple of different unique features in our solution that are never before seen in maritime domain. One of such is our sensor unit, in our sensor unit we have integrated into one unit two different cameras 4K day camera with 225 degrees field of view and thermal with 180 degrees field of view, double GNSS positioning sensors, Inertial Measurement Unit and in the unit, there is also integrated AIS receiver and we can bring in the ARPA detections from external radars as well. Also with all these different data sources we do sensor fusion that makes interpreting the information for user much easier.
Then another unique feature is the where one can see the day video and thermal video on top of each other.
Sensor fusion means that we take information from multiple sources and put them together. Many information sources are good at something but then they are not so good on something else. With sensor fusion we try to utilise the best parts of all information sources. For example, AIS is good for providing information on large vessels around you, but the information is not that accurate, then to make this accurate we use our machine vision detect different vessels and we match the detections to AIS signals and this way we can provide accurate location information to compliment the AIS information.
There are actually three different cameras inside the sensor unit for the day camera and then also three for the thermal camera. The way the object detection works is that we have for more than two years now collected images of different vessels in different conditions. Then we tell our machine learning system what is in that picture, for example if there is a container ship, fishing vessel or a tugboat. We also tell which direction it is heading. Then we use this information to teach the machine vision model so that next time the camera sees something similar it can determine what it is. This teaching process takes time, and it is never perfect, therefore we will be collecting image information also for years to come in order to make our system more accurate.
We can classify between the following vessel types;
Yes we can in theory, however we will have to train our models for machine learning works in a way that it can detect what it is trained to detect. In order to train the detection with new objects the required image data for one new item type is around 10 000 images in different weather conditions and from different viewing angles. We can train customized detection types if customer has requirement for detecting some special type of objects.
As non vessel type detections currently Groke Pro detects for example sea markers.
The range always depends on multiple factors like weather, object size and such. But as reference we can say that our machine vision has detected for example a cargo vessel which was 2.5 nautical miles away from our sensor unit.
Thermal camera provides you with thermal difference view. This means that for example in night-time when using thermal view one can more easily see thermal sources around you. The thermal image is usually monochromic but in our solution, we can bring in colours in the blended view from our day camera.
No, the system utilises also day and thermal camera for detections which means with cameras we can detect also non-AIS targets and we can also bring in detections from radar (optional feature).
Yes, we have built in standard interface towards some radar models already and can add new ones if needed. The settings for radar detections are still done at radar, we just overlay them into our sensor fusion. This is an optional add-on feature to our Groke Pro solution.
Yes, the system is a stand-alone system which functions without any connectivity. However we do provide a modem that includes a SIM card, so when there is land connectivity, it will automatically upload the images/data taken for teaching our machine vision system, through that connection we also provide software updates to our system if needed and update the machine vision detection models as well.
The system can have three user interface devices maximum, which will always include one tablet. The remaining two could be either an additional tablet and/or a fixed monitor.
The system will be only work where it is in reach of Wi-Fi from the central unit, and the assumption is that it can only be used in the bridge. However, we can provide as optional add-on extra Wi-Fi so that tablet can be used elsewhere in the vessel.
The installation of our system is straightforward because there are only few components to be installed. Sensor unit has its own mounting bracket and mounting stand depending on the installation location, then we have the central unit which shall be mounted near bridge into an air conditioned location. Then we have the user interface device and its connectivity and that is it. From cabling side, the system only needs couple of Ethernet cables and power supply. Installation will take totally around 4-6 hrs.
The preferred location for sensor unit is on top of the compass deck, but it can be installed also to mast. The central unit shall be installed somewhere inside the vessel that will not be exposed to outside air.
Of Crouse, all of our installations so far are retrofit installations and the system is designed in a manner that it is easy to put into existing vessel and it does not take too much space.
Yes, installation is easy and straightforward for new building vessels. Also, with new buildings it is easier for example to go with our roof display that enables information sharing between bridge crew members because in the building phase it is easier to design the display into the roof elements.
As our system is advisory system and not navigation system, it does not need type approval but we have selected marine type approved components as far as practically possible, for some components like the high performance computers there are no marine approved version available so we are using industrial use graded components in these cases. The system is tested according to relevant standards.
We are also working closely with class societies constantly.
We have warranty for 1 year for the product from commissioning.