Co-founder @zipline, using fleets of robot airplanes to approximate teleportation

San Francisco
when Ziplines need maintenance, they fly themselves home. Increasingly, they fly home by hopping across our network of charging sites. Today in Houston:
10
6
183
8,239
Ryan Oksenhorn retweeted
Replying to @ryanzip
@ryanzip amazing service as always 🙌🏼 I am taking advantage of the free delivery fee while I can👍🏼👍🏼
1
16
571
and for good measure, we hit it with a second zap. bigger bang!
Video results of zapping a zipline
3
6
71
4,706
Video results of zapping a zipline
the things you see on slack at this company. never a dull moment at zipline
2
3
84
11,684
the scale in this video from last year still blows my mind, but what's even wilder is that since filming this we've changed the system to deliver from >50% higher than what you see here in our obsessive quest to be silent & invisible
7
5
82
2,986
the things you see on slack at this company. never a dull moment at zipline
6
5
133
18,210
It only took 30 years to notice this, but there is a glaring issue in the PokeRap s01e03
3
13
1,472
how it started how its going
29
39
586
42,491
More cool robotics demos today, but remember the Brittle Lesson: mechanical complexity is brittle in the real world. How many hours of real-world use will that hardware get before one of the 10k+ parts break?
Robotics companies have all accepted The Bitter Lesson and we now see awesome demos, but most have no idea about The Brittle Lesson and boy are they in for a surprise when trying to productize.
3
1
46
2,984
The Brittle Lesson is coming for many robotics companies that will soon need to move from flashy demo → reliable product
Robotics companies have all accepted The Bitter Lesson and we now see awesome demos, but most have no idea about The Brittle Lesson and boy are they in for a surprise when trying to productize.
2
43
2,807
Robotics companies have all accepted The Bitter Lesson and we now see awesome demos, but most have no idea about The Brittle Lesson and boy are they in for a surprise when trying to productize.
23
28
403
44,670
Hardware is hard. Every sensor, every part, every degree of freedom is a burden. The more parts, the more failure points. That brittleness is the enemy, but this brittle lesson clearly hasn't been learned by most teams. Instead, you see them touting high DOF as a feat!
1
76
3,080
A better heuristic for which robotics co's will succeed is: study how much work went into minimizing DOF, part count, and # of sensors.
1
1
72
2,801
90th birthday cake, delivered.
9
6
111
3,855
it's a beautiful day for flying
3
2
134
2,491
Zipline's footprint in Africa is growing faster than at any point in our nearly ten years of operations. Exciting updates coming soon!
10
9
157
3,830
You're not using enough AI and it is your fault when LA floods.
Los Angeles, CA (moments ago) Local resident of 30 years has never seen the LA river reach this level following a series of thunderstorms that swept through the region this afternoon.
3
2
72
7,070
seeing these wild Astra blender examples is bringing back this thought: Astra actually hits orbital velocity. It can stay up in orbit seemingly indefinitely, without decaying on its task over time.
Why things suddenly feel different after gradual progress: AI reached orbital velocity. There is no value in a velocity 1m/s slower than orbital. But once you gain 1m/s....
1
1
13
2,477
.@zipline now flies in Sport Mode
Replying to @ryanzip
Dynamic Flight Trajectories are like hyper-miling: realtime modelling of wind, airspeed, lift, and drag optimizes speed + altitude profile to minimize time, energy, and noise. Each Zipline refines its plan in flight as conditions change. We now climb while moving toward destination, descend to build airspeed, climb to slow down. We let gravity do more oft he hard work.
3
5
117
7,883