Design and experimental testing of a robotics control architecture for free-running amphibious models
Michael Swafford, Casey M. Harwood, Pablo M. Carrica, Venanzio Cichella · 2020
To date, advancement in autonomous vehicles has been substantial, from Tesla's Full Self-Driving beta released in October of 2020 to semi-autonomous surface vessels like the U.S Navy's sea hunter; autonomy is the future of industry and commercial products. While both of these examples are significant accomplishments, both are limited to single-mode vehicles. Multi-modal vehicles like sea-going autonomous amphibious vehicles (AAVs) must replicate the capabilities of autonomous land and water vehicles, but must also transition between them in an environment, such as the littoral surf zone, that is arguably more complex than that encountered by either of the preceding examples. In order to successfully design a vehicle and its control systems to operate in this environment a thorough experimental and computational study of a model AAV must be conducted in order to understand both the vehicle and the environment it will serve in. Presented in this paper is the design and development of an robotics control architecture for free-running amphibious models with the capability to expand to full autonomy. The architecture was used to fulfill a second phase of experimental testing for the Model Quadski (MQS) in an artificial surf zone at the IIHR wave basin facility. The architecture uses a land based robot operating system (ROS) PC to communicate with both the model and other computers in the facility. For this phase of experimental testing the ROS package utilized a feed-forward control scheme in a conjunction with an XBOX controller to semi-autonomous control the robot. It was shown that the upgrades to the control authority of the MQS enabled a free-running model amphibian that could drive on land, in water, and seamlessly transition between the two. The upgrades to the hardware and software of the vehicle were tested at the IIHR wave basin facility in both calm water and surf conditions. The ROS package that was developed added a clear advantage to the complexity of experiments that could be undertaken in the surf zone and enabled the collection of high quality data of vehicle transit through the surf zone for both launching and landing cases. The beginning development for closed-loop control has also been demonstrated by the ROS package taking information from the real-time motion tracking system and feeding it into a Simulink node where a closed-loop control scheme could be added in future phases of the project.