Autonomous boats ferrying people and goods, and helping prevent road congestion, could be the next-generation transportation in waterway-rich cities such as Venice, Bangkok, and Amsterdam, where canals run next to and below busy bridges and streets.
Image credit: MIT
MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) and the Senseable City Lab in the Department of Urban Studies and Planning (DUSP) have made attempts to realize this by developing a fleet of autonomous boats that enable precise control and high maneuverability. It is also possible to rapidly 3D-print the boats with the help of a low-cost printer, increasing the feasibility of mass manufacturing.
The boats can even be employed to ferry people around and to transport goods, thereby reducing street traffic. The researchers also hope that in the future, the driverless boats could be used to provide city services overnight, rather than during the bustling hours of the day, which will further reduce the congestion on roads as well as canals.
Imagine shifting some of infrastructure services that usually take place during the day on the road - deliveries, garbage management, waste management—to the middle of the night, on the water, using a fleet of autonomous boats,” stated CSAIL Director Daniela Rus, co-author on a paper, depicting the technology that will be presented at the IEEE International Conference on Robotics and Automation this week.
The boats include rectangular hulls measuring 4m x 2 m and are fitted with GPS modules, microcontrollers, sensors, and other hardware. Therefore, they can be programmed to self-assemble into platforms for food markets, concert stages, floating bridges, and other structures within a few hours. “
Again, some of the activities that are usually taking place on land, and that cause disturbance in how the city moves, can be done on a temporary basis on the water,” stated Rus, who is the Andrew and Erna Viterbi Professor of Electrical Engineering and Computer Science.
Environmental sensors can also be fitted in the boats to monitor waters of a city and to understand urban and human health.
Co-authors of the paper are first author Wei Wang, a joint postdoc in CSAIL and the Senseable City Lab; Luis A. Mateos and Shinkyu Park, both DUSP postdocs; Pietro Leoni, a research fellow, and Fábio Duarte, a research scientist, both in DUSP and the Senseable City Lab; Banti Gheneti, a graduate student in the Department of Electrical Engineering and Computer Science; and Carlo Ratti, a principal investigator and professor of the practice in the DUSP and director of the MIT Senseable City Lab.
Better Design and Control
The study was performed as part of the “Roboat” project, a partnership between the MIT Senseable City Lab and the Amsterdam Institute for Advanced Metropolitan Solutions (AMS). As part of the project, in 2016, the team tested a prototype that ferried around the canals of the city’s, by moving it forward, backward, and sideways about a preplanned path.
The ICRA paper describes various significant new innovations: a highly efficient and agile design, a fast fabrication method, and sophisticated trajectory-tracking algorithms that enhance control, precision latching and docking, and other tasks.
The researchers built the boats by 3D-printing a rectangular hull using a commercial printer, making 16 separate sections that were bonded together. It took nearly 60 hours to perform the printing. Then, multiple fiberglass layers were adhered to seal the completed hull.
GPS, power supply, a minicomputer and microcontroller, and Wi-Fi antenna are integrated onto the hull. To enable precise positioning, the researchers fitted an indoor ultrasound beacon system and outdoor real-time kinematic GPS modules, which enable centimeter-level localization, and also an inertial measurement unit (IMU) module that monitors the angular velocity and yaw of the boat, apart from other metrics.
In contrast to the conventional catamaran or kayak shapes, the boat is shaped to be rectangular to enable the vessel to move laterally and to fix itself to other boats during the assembly of other structures. Another simple but effective design element was the placement of four thrusters at the center of each side, rather than at four corners, thereby producing forward and backward forces. The researchers stated that this renders the boat more agile and efficient.
The researchers also devised a technique for allowing the boat to track its position and orientation in a more rapid and accurate manner. To achieve this, they created an efficient version of a nonlinear model predictive control (NMPC) algorithm, which is normally used for the control and navigation of robots within various constraints.
Algorithms such as the NMPC and other similar ones have been employed even earlier to control autonomous boats. However, essentially, those algorithms are tested only under simulation conditions and do not account for the boat’s dynamics. In contrast, the team included simplified nonlinear mathematical models in the algorithm that account for a few familiar parameters, such as centrifugal and Coriolis forces, drag of the boat, and mass added because of acceleration or deceleration in water. An identification algorithm was also used by the team, which then detects any unfamiliar parameters when the boat is trained on a path.
Finally, an efficient predictive-control platform was used by the team to run their algorithm. The platform could quickly detect upcoming actions and increased the speed of the algorithm by two orders of magnitude over similar systems. When compared to other algorithms that executed within around 100 ms, the algorithm developed by the team took less than 1 ms.
Testing the Waters
To exhibit the efficiency of the control algorithm, the team operated a smaller prototype of the boat along preprogrammed paths in the Charles River as well as in a swimming pool. During the 10 test runs, the team observed average tracking errors, in orientation and positioning, which were smaller when compared to tracking errors of conventional control algorithms.
This accuracy was achieved partly due to onboard GPS and IMU modules in the boat, which respectively identify the position and direction, down to the centimeter. The NMPC algorithm analyzes the data from the modules and considers different metrics to steer the boat in real-time. The algorithm, which is executed in a controller computer, individually controls each thruster, updating every 0.2 seconds.
The controller considers the boat dynamics, current state of the boat, thrust constraints, and reference position for the coming several seconds, to optimize how the boat drives on the path. We can then find optimal force for the thrusters that can take the boat back to the path and minimize errors.
Wei Wang, Co-Author
According to the researchers, the innovations in design and fabrication, and also rapid and more precise control algorithms, indicate that it could be possible to use driverless boats for docking, transportation, and self-assembling into platforms.
The next stage in this study would be to develop adaptive controllers to account for variations in drag and mass of the boat while transporting goods and people. The team has also been striving to fine-tune the controller to account for stronger currents and wave disturbances.
We actually found that the Charles River has much more current than in the canals in Amsterdam,” stated Wang. “ But there will be a lot of boats moving around, and big boats will bring big currents, so we still have to consider this.”
A grant from AMS supported the study.
Video credit: MIT