Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:03):
Drone racing is a thrilling competition that pits talented pilots
against each other in a contest of skill and daring,
and now robots are getting in on the action. I'm
Jonathan Strickland and this is text Up Daily. NASA's Jet
Propulsion Laboratory in Pasadena, California has lots of stuff going on.
(00:25):
Among the many projects that the numerous actual real life
rockets scientists have initiated is one that teaches an artificial
intelligence how to pilot a small drone through a complex
obstacle race course, which frankly sounds like the coolest job ever.
The research team working on the project received funding from
(00:46):
Google and took two years to create the software that
could learn a course and send the appropriate commands to
a drone to make its way through without crashing or
missing a turn. The team built three drones called Batman, Nightwing,
and Joker. The drones have a top speed of eighty
miles per hour or a d kilometers per hour if
they have enough space to get up to top speed
(01:09):
within the confines of the race course, that top speed
is closer to forty miles per hour or sixty four
kilometers per hour. As a test of their work, the
research team invited a professional drone racing pilot, and yes,
in case you weren't aware, that's a thing that really exists.
This pilot was Ken Lu, a k A Flying Bear
(01:31):
on the drone racing circuit, who averaged a blistering pace
of eleven point one seconds per lap on the course.
The AI piloted drone was just a touch slower, with
an average time of thirteen point nine seconds. So for now,
a top performing human can still outpace an AI piloted drone,
But Lou himself was impressed by the performance. Lu said
(01:54):
to reporters that quote one of my faults as a
pilot is I get tired easily. When I it mentally fatigued,
I start to get lost, even if I've flown the
course ten times end quote. That's an advantage AI has
over humans. It doesn't get sleepy or distracted. It can
continue to operate as long as it has the computational
(02:15):
power necessary to execute commands. The AI s path was
also more consistent than LOSE technique for navigating the course.
While LOSE drone might vary in its flight path with
each lap, the AI piloted drone continued to follow the
same line time after time. The technology has the potential
to benefit multiple industries. Space exploration is definitely high on
(02:37):
that list. The vast distances involved in space exploration make
it impossible to rely upon a human operator controlling a
remote vehicle. There's too much delay between when a person
would send a command and when that command would reach
the robot. When the Curiosity rover made a soft landing
on Mars, it did so autonomously because there was a
communication delay of several minutes due to the distance between
(02:59):
Earth and Mars. The rover was actually on Mars for
about ten minutes before scientists on Earth could confirm touchdown.
Autonomous systems are critical for situations in which a vehicle
must navigate far from human control back here on Earth.
A versatile navigation and mapping system could help a drone
become part of an automated operations and everything from search
and rescue missions to something as simple as checking on
(03:22):
inventory inside a massive warehouse. There are numerous potential benefits
to AI controlled drones. As for drone racing, that's a
sport that's literally taking off. Numerous hobby organizations host drone
racing competitions, and some pilots compete to set the best
course time, navigating through specific obstacles and flying drones through
(03:42):
gateways or around pylons and others. Multiple pilots might compete simultaneously,
racing against each other to get to the finish line first.
Pilots might even don a head mounted display to view
a live video feed footage from a camera on the
drone to fly the course from a first person first effective.
The big name in drone racing is d r one Racing,
(04:04):
but there are many others. Perhaps in the future will
see some artificially intelligent competitors enter the field. It's not
difficult to imagine a scenario in which machines are able
to outperform even the best human pilots. In some ways,
that's a sad realization, but we should also remember that
humans were the ones able to create the machines in
the first place, and that they have the potential to
(04:25):
do some phenomenal things beyond going really fast through an
obstacle course. That's all for today. To learn more about drones,
artificial intelligence, and NASA, subscribe to the Tech Stuff podcast.
I cover topics like this and other important concepts. In
tech and in much greater detail. New episodes published on
Wednesdays and Fridays. I'll see you again soon.