难题是，善于一种交通出行方法的智能机器人一般 不善于另一种。机载无人飞机迅速灵便，但一般 电池循环次数比较有限，没法开展远距离航空。另一方面，路面车子的能耗等级更高，但速率变慢，操控性更差。
Walking and flying is a typical characteristic of nature-many birds, insects and other small animals can guarantee this aspect. If people can make intelligent robots with similar multifunctionality, it will open up a lot of possibilities: Imagine that the device can fly into the engineering construction area or the seismic area not close to the road, and then squeeze through the narrow indoor space on the road to transport objects Block or rescue.
The problem is that intelligent robots that are good at one method of transportation are generally not good at another. Airborne unmanned aircraft is fast and convenient, but the number of battery cycles is generally limited and it is impossible to carry out long-distance aviation. On the other hand, road vehicles have higher energy consumption levels, but their speeds are slower and their handling is worse.
Scientific research staff from the Massachusetts Institute of Technology's Electronic Information Science and Artificial Intelligence Technology Laboratory (CSAIL) is focusing on the development and design of intelligent robots that can maneuver on land and fly to the sky. In a new graduation thesis, the team showed a system software consisting of 8 quadcopters. The system software is capable of aeronautical and safe driving in a natural environment close to a big city. It also has a parking lot, no-flying Zone and helipad.
The first author of this graduation thesis, PhD student Brandon Araki, said that the working ability of aviation and safe driving is effective in a natural environment with many obstacles, because you can pass over road obstacles You can also drive safely under obstacles on your head. General drones cannot maneuver on the road. A wheeled drone is more manoeuvrable, and aviation time can only be reduced slightly.
Araki and CSAIL principals Daniela Rus, and their MIT undergraduates JohnStrang, Sarah Pohorecky, CelineQiu, and Tobias Naegeli of the Federal Government Institute of Technology Zurich's Advanced Interaction Laboratory developed this system software.
How does this work
The new project was created based on Araki's previous work. Basically, a "flying little monkey" intelligent robot was developed and designed to crawl, crawl and fly. Although the little monkey intelligent robot can bypass obstacles and crawl around, it still cannot move independently.
In order to deal with this problem, the elite team developed and designed a variety of optimization algorithms and optimization algorithms, dedicated to ensuring that UAVs are not easy to collide with. In order to make them safe to drive, the scientific research team installed two small motors with wheels on the bottom of each drone. In the simulation, this intelligent robot can fly 90 meters or safely drive 252 meters before the rechargeable battery runs out.
After adding driver components to the drone, its battery cycles have been slightly reduced, which means that its larger aerospace has been reduced by 14% to 300 inches. However, because safe driving is still much more efficient than aviation, the increase in safe driving efficiency has filled the relatively small damage to aviation high efficiency caused by overload.
Yu Jingjin, a professor of electronic information science at Rutgers University, said that in this work, an optimized algorithm solution was presented for large-scale, mixed-mode transportation, and the information showed that it was acceptable for practical problems. Sex. Yu Jingjin did not participate in this scientific research.
The elite team also tested the system software using usual raw materials, such as textiles for pavement and cardboard boxes for engineering construction. They tested eight intelligent robots from the starting point to the end station on a collision-free relative path, and all of them were successful.
Rose said that system software like them shows that another way to make a safe flying car is not to simply install an aircraft wing on the car, but to basically improve the safety of unmanned aircraft in many years of scientific research Driving ability.
When you first started developing the overall planning and control system for flying cars, people were inspired by the probability of creating intelligent robots with this capability in a small area, Ross said. Although it is obvious that it will continue to undergo great trials to expand to a real passenger-carrying vehicle, flying cars will show people a rapid and traffic-free vehicle in the future, and this development potential has inspired people.