Drone racing is an increasingly popular sport offering significant bonuses to qualified professionals. New control algorithms developed by the University of Zurich (UZH) defeated experienced human pilots for the first time, but they still have significant limitations.
In the past, attempts were made to develop automated algorithms to defeat humans, but encountered problems in accurately simulating the limitations of quadcopters and their flight paths. The traditional flight path around the complex drone race track is calculated using polynomial methods, which produce a series of smooth curves that are not necessarily as fast as the sharper, more irregular paths that human pilots fly. A team from the
UZH Robotics and Perception Group has developed a trajectory planning algorithm to calculate the best route for each point in flight, rather than segment by segment. The previous
UZH autopilot algorithm has been shown to be effective in avoiding obstacles and passing through doors that make up the route, but the speed is significantly slower than that of inexperienced drone pilots. The result this time is better.
“Our drone exceeded the fastest lap of two world-class human pilots on the experimental track,” said team leader Davide Scaramuzza.
In addition to being faster, the automation system is more consistent. Humans can have good and bad circuits, but machines always offer the best games and the probability of accidents is much lower.
is more suitable for you
Russia has a new stealth fighter: what is it for?
The US shipbuilding industry is at its lowest point in history. How did the United States fall to this day?
target: Taichung. This port in Taiwan may be China’s main target.
However, the success of this experiment comes at a price. One is that the flight plan requires more computing power than the drone’s onboard computer. The current plan for a few minutes of flight takes up to an hour. Managing flights in real time will require more efficient algorithms or better drone computers.
“We believe that online re-planning can and will soon be possible,” said researcher Philipp Föhn, the first author of the new paper. “There is a trade-off between optimization and re-planning enough to defeat human pilots.”
Racing drone 3D capture
UZG AIs racign drone test was also placed on a huge motion capture stage to track… . [+] Another limitation of UZH
is that the position of the drone must be monitored by a series of external cameras so that it can accurately track its progress along the planned flight path in 3D; this requires one of the largest motion capture systems in the world. A practical solution should only apply to data collected from the drone’s own sensors.
This work was funded by the Swiss National Research Competence Center as part of its rescue robot challenge, which aims to develop automatic search and rescue technology and allow drones to find victims faster. The flight planning system that can produce the fastest safe route can also bring real benefits to various other aircraft.
“For other aircraft or general 3D motion planning, multiple goals must be met in sequence, and our method may be a feasible solution,” Föhn said.
This may include urban air taxis and package delivery, as well as efficient inspections of infrastructure and other emergency services by drones.
In a way, this victory is similar to last year’s AlphaDogfight test (or a recent Chinese copy), when artificial intelligence pilots decisively defeated humans in a simulated aerial battle. The difference is that the drone competition takes place in the real world. It’s one thing to beat humans in computer games, but in drone competitions, they can now beat us in the real world. Prepare to witness a new generation of machines that are truly superhuman.
Follow me on Twitter. Check out my website or some of my other works here.
David Hambling
David Hambling
Author of “Swarm Troopers: How Small Drones will Conquer the World”, following the general cutting-edge military technology, especially unmanned systems. New science…read more
reprint and license
June 30, 2021, 10:18 am EST | 106,019 views
planning for a mixed work world
Jason Girzadas Brand Contributor
DeloitteBRANDVOICE | Salary Plan
As the line between work and family blurs, companies are looking for various models to keep the team engaged. Although more and more organizations bring employees to the office after working from home for more than a year, other organizations have decided to continue to allow people to work remotely. Then there’s the hybrid approach, a new way of working to support future jobs that existed long before the pandemic, which continues to provide workers with the ongoing flexibility and need for face-to-face collaboration and connection they crave. For the company, this is a real-time opportunity to learn how to cultivate a culture that is remote and close at hand.
people stand on the mountain
Maintaining the organizational culture has never been more challenging. People feel isolated and the lack of impromptu interaction means that opportunities for integration and idea generation must have different styles. However, the pandemic has also tested policies, technological advances and investments that make collaborative work possible without face-to-face interaction. Our Deloitte analysis shows that in the first eight months of 2020, at least 100 digital remote collaboration tools have been brought to market or enhanced. According to a recent workplace survey of 275 executives, 68% of people want to implement a hybrid mode of operation. Physical and virtual work.
When people do not share physical space, they still need effective forms of cooperation. When transitioning to mixed work, leaders cannot ignore the value of high-level collaboration, and there are some practical ways to prepare teams and organizations for this transition.

By Peter

Leave a Reply

Your email address will not be published.