Recently, the official website of MIT reported the newest achievement in scientific research- unmanned surface ships of Wei Wang who graduated from College of Engineering of Peking University (news link: http://news.mit.edu/2018/fleet-autonomous-boats-service-cities-reducing-road-traffic-0523). This project has been funded by AMS. Relative dissertation has been issued in ICRA conference and Wei Wang is the first author.

The content about the news:

Highly mobile unmanned ship, aiming for mitigating traffic congestion of the city with numerous waterways

In water-rich cities such as Amsterdam, Bangkok and Venice, the means of transportation in the future will probably be self-driving ships, which can be used to transport goods and people and help clear up road congestion.

The Technology Computer Science Lab of Massachusetts Institute，CSAIL and DUSP are planning a series of highly automated and precisely controlled autonomous boats, which means a great progress in this field. Take a step. Use a low-cost printer to quickly print a boat in 3D, making mass production possible.

These boats can be used to transport goods and people to ease traffic congestion in the streets. In the future, the researchers also imagine that the unmanned boat could also play an important role in city services at night instead of working at the busy daytime to further ease the congestion of roads and canals.

“Imagine that some infrastructure services that are usually carried out during the day, such as freight, garbage management, and waste management, are changed to use a fleet of automated ships on the water in the middle of the night.” Co-authors of the paper, CSAIL Director, Andrew and Wiener Daniela Rus who is the professor of electrical engineering and computer science, will describe the IEEE robot and automation in International Conference in this week.

In addition, the vessel is equipped with sensors, microcontrollers, GPS modules and other hardware. It can program a rectangular hull with 4 meters long and 2 meters wide, making it self-assembled into a pontoon, concert stage, food market platform and other structures, which can maintain several hours. “Some activities that occurring on the land may interfere with the traffic in the city can be carried out temporarily on the water,” said Rus.

The ship can also be equipped with environmental sensors to monitor the water condition and observe the health of cities and humans.

　　

 

Better design and control

This technology is part of the "Roboat" project which is a joint project of the MIT Senseable City Lab and the Amsterdam Advanced Urban Solutions Institute (AMS). In 2016, as part of the project, the researchers tested a prototype boat cruising on a city canal, moving forwards, backwards, and laterally along pre-designed paths. To make the ship, the researchers printed a rectangular hull using a commercial printer 3D, producing 16 spliced separate parts. This process took about 60 hours. The finished hull is then sealed by several layers of fiberglass, and the power supply, Wi-Fi antenna, GPS, and small computers and microcontrollers are integrated into the hull. For precise positioning, the researchers also add the indoor ultrasonic beacon system and outdoor real-time dynamic GPS module (allowing centimeter-level positioning) and other indicators of the inertial measurement unit (IMU) module to measure the yaw angle and angular velocity of the ship. The shape of the boat is a rectangle rather than a traditional kayak or catamaran shape, which allows the boat to move laterally and attach itself to other boats. Another simple and effective design element is thruster placement. The four thrusters are located in the center of each side instead of the four corners, producing fore and after forces. The researchers said that this design makes the ship more flexible and efficient. The team also developed a method that enables the ship to track its position and orientation faster and more accurately. For this purpose, they developed an efficient version of the Nonlinear Model Predictive Control (NMPC) algorithm, which is commonly used to control and navigate robots under various constraints.

NMPC and similar algorithms have been used to control autonomous boats. But in general these algorithms are only tested in simulation environment or do not consider the dynamics of the ship. The researchers added simplified nonlinear mathematical models to the algorithm that takes some known parameters into account, such as ship resistance, centrifugal force, and Coriolis force, as well as the quantity increase caused by water acceleration or deceleration. The researchers also used a recognition algorithm that identified any unknown parameters that the vessel was trained on the path.

Finally, the researchers used an efficient predictive control platform to run their algorithms. The algorithm can quickly determine the upcoming action. The algorithm is two orders of magnitude faster than similar systems. Although other algorithms are performed in about 100 milliseconds, the researchers' algorithms require less than 1 millisecond.

　　

Water Test

To demonstrate the effectiveness of the control algorithm, the researchers deployed a smaller boat prototype along the pre-planned path in the pool and Charles River. During the 10 tests, the researchers observed that the average tracking error in positioning and orientation was less than the tracking error of traditional control algorithms.

The increase in accuracy is due to the GPS and IMU modules on board, which determine the position and direction, and can be accurate to centimeter. The NMPC algorithm processes the data of these modules and measures various indicators to guide the ship how to work. The algorithm is implemented in the controller computer and adjusts each thruster individually, updated every 0.2 seconds.

"The controller will take into account the ship's dynamics, the current state of the ship, the thrust constraints, and the reference position for the next few seconds to optimize the way the vessel is driving on the road," Wang Wei said. "Then we can find the best propeller Power can push ships back on the road and minimize errors."

Researchers say that innovations in design and manufacturing, as well as faster and more precise control algorithms, point to viable autopilot ships for transport, docking and self-assembly into platforms.

The next step in this work is to develop adaptive controllers to solve the changes in the quality and resistance of the ship when transporting people and cargo. The researchers are also improving the controller to deal with wave disturbances and stronger water flow.

“We actually found that the Charles River's water flow is more complicated than the Amsterdam canal,” Wang Wei said. “But there will be many boats moving around in Amsterdam. Large ships will cause large currents, so we still need to consider this.”

　　

ote:Wei Wang was the master-doctoral student of Professor Guangming Xie of the College of Engineering of Peking University. The doctoral thesis works mainly on autonomous intelligent bionic robotic fish. After his graduation in 2016, he went to the Massachusetts Institute of Technology as a postdoc.

 

COE News , PKU：http://www.coe.pku.edu.cn/news-express/5434