As the SeaClear project progresses, Fraunhofer CML has been busy with the design and development of the collection basket for the autonomous system. The basket is designated to aggregate collected litter and ensure a safe transport out of the water and later on to shore, at the end of each mission. A modular prototype was built that allows for experiments with different wall materials, ROV-to-Basket docking areas, and Gripper-to-Basket interfaces.
On the 15th and 16th of September 2021, the first tests of the SeaClear robots took place in the Dubrovnik area, Croatia. This was the first time representatives from all the partners met live after the kick-off meeting of the project, due to the on-going pandemic due to CoVID-19.
As we are approaching the final trials for the SEACLEAR system, our team has gathered once again in Marseille from 27 of February until the 3rd of March for further tests. Updated versions of the components were tested alone and in combination with each other, to discover improvement points.
An updated version of the basket was brought on site, featuring AR tags for easier localisation from the Tortuga, fiber brisstles for easier deposit of litter without possibility of escape, and a funnel shaped entry for easier entrance of the Tortuga with the gripper. The new basket was also tested in tandem with its LARS module (Launch And Recovery System) that was developed for the SeaCAT.
Furthermore, the Tortuga was equipped with a refined version of the gripper, which was 3D printed in aluminium to optimize for weight and resistance. Several tests demonstrated that the new gripper can handle litter of different shapes and dimensions, and is ready for deployment for the final demonstrations. Finally, we also managed for the first time to perform real-time localisation of the tortuga using the drone that was flying above.
We are now on the final stretch before our much anticipated demonstrations on Hamburg this June. Stay tuned for updates and details about the project results!
The “eyes and ears” of SeaClear robotic platform, aka the observation ROV, has been going through a development phase in order to accurately perceive its surrounding environment. Different computer vision tasks were examined such as instance segmentation and object detection in order to select the one that best satisfies the SeaClear requirements and limitations in terms of accuracy and computational cost. Similarly to all neural network architectures, convolutional neural networks (CNNs) that are employed in this computer vision task require big amounts of data. Hence, additional data have been collected during the SeaClear pilot tests, which were afterwards processed and labelled. In total, we have generated 8610 labelled image samples by the measurements conducted during the SeaClear pilot.
After generating a sufficiently large dataset, different convolutional neural network architectures have been examined. We have explored both two-stage and one-stage CNNs showing promising results. Two-stage CNNs generate regions of interest and use these region proposals to perform object classification and bounding box regression. In contrast, one-stage CNNs perform object detection in one stage, ensuring much faster inference.
Classification example in a controlled environment
Next, we want to focus on optimizing these CNN architectures and testing them in real time. Stay tuned for more updates to come.