To make piloting ROVs easier, researchers at the University of Hawai’i created an augmented/virtual reality interface and a hybrid autopilot system. They modified the software of the BlueROV2 to improve its control and added features that allow the vehicle to have varying levels of autonomy. This system was tested using a simulation and real-world water tank experiments. The results showed a significant improvement in the robot’s behavior prediction.
Abstract: Remotely operated vehicles (ROVs) are marine submersible robots that serve a variety of purposes in industry and research. These unmanned vessels harness human judgment and precision to perform tasks within extreme environments like the deep sea, polar, and volcanic regions of the ocean. Some examples of their usages are to survey the ocean floor, maintain pipelines and collect scientific data in the form of sediment and hydrothermal vent plume samples and optical observations of marine wildlife. Training of ROV pilots is typically very expensive and time-consuming because of the highly specialized skill requirements. A novel system was proposed by collaborators from the University of Florida (UF) and the University of Hawai‘i at M¯anoa (UHM) for piloting ROVs with an intuitive augmented/virtual reality (AR/VR) interface that uses a hybrid autopilot. To demonstrate the feasibility of this system the group at UHM altered the ArduSub firmware of the commercially available BlueROV2 (BROV2) Heavy to enable model-based quantitative control. Error feedback control for the hybrid autopilot was implemented using Robot Operating System (ROS) in a modular manner that enabled various levels of autonomy to assist ROV pilots. Alongside the development of the custom firmware and hybrid autopilot, the software-in-the-loop (SITL) simulation environment was also updated with an experimentally determined hydrodynamic model using onboard sensorbased system identification techniques. As much as sixty percent improvement of relative error when predicting vehicle behavior compared to the original SITL model. The calibration process of the hybrid autopilot involved iteratively cycling between SITL and water tank testing. It was demonstrated that this procedure was an effective method for achieving precision control of the BROV2.
Author: Ng, P.
Journal: University of Hawai’i at Manoa
To read more, click here!