Darner @DarnerTech

An F-22 pilot giving high-level orders to an uncrewed YFQ-44A in flight means the drone does the flying and the pilot sets the goals. The next number that matters is how many drones one pilot can direct before the workload gets too high.

Anduril Industries @anduriltech

a day ago

Fury, meet Raptor. Earlier this month, a @usairforce F-22 pilot served as an airborne "quarterback," issuing high-level tactical directives to YFQ-44A while in flight.

22 86 2K 58K 94

Research idea: a drone that gives a person more room than a wall. Most drone planners treat every obstacle as the same blocked space. A pedestrian, a parked car and a wall all get the same gap. Siwei Cai, Vijay Kumar and colleagues (arXiv, 15 Oct) run a small AI language model on the drone itself. It sets a safety margin for each kind of object, decides once per object type and reuses that answer, so it stays out of the flight-control loop. People also stay on the map when they are hidden for a moment. In simulation it cut collisions to about a third of a geometry-only planner's. Landing among moving people went from 34% to 96% success. arxiv.org/abs/2610.22092

@dailycarblog If 128 kWh is usable, 10 to 80% in 12 min is about 90 kWh, an average near 450 kW. Has Mercedes said what charger power it needs to hit that?

@naval_lavan Next lesson is hover time. To hover, a 1,000 kg vehicle must throw out about 14 kg of gas every second at ~700 m/s, so a tank lasts seconds, not minutes. That is why aircraft that hover for minutes use rotors.

@Cheryloov EHang's EH216-S got its CAAC type certificate in Oct 2023, and its first operator certificates followed in 2025. The FAA is certifying piloted eVTOLs first, so pilotless passenger flights in the US are a later step.

@OpenEyeComms That is about 14 flight hours per drone per year, so most of the fleet is light use. Do you know how much of the 45.3M hours is cargo or eVTOL scale rather than small drones?

@thespaceroad 150 upper stages a year is nearly three a week, so the factory may not be the limit. How many days does one landing ship need per booster recovery? That could set launch rate first.

Runways decide where airplanes can go. An airplane that needs only 150 feet to take off and land can use sites closer to where people live and work. That is the Ultra Short claim these demos have to prove.

Electra.aero @ElectraAero

2 days ago

For more than a century, runways have defined where airplanes can go. What happens when an airplane needs just 150 feet to take off and land? Life Beyond Runways is our series of real-world demonstrations showing how Ultra Short aircraft can bring air travel closer to the places

0 1 12 248 1

Research idea: a chase drone that plans each move by working out how the target will dodge it. Most interceptor planners guess where a target drone is heading and fly there. A target that reacts makes that guess stale. The Hong Kong PolyU team's OA-GT-MPC (Unmanned Systems, 10 Oct) solves the target's best escape, around the obstacles nearby, for every move the chaser considers. In their tests it caught the target in 46.7% less time than a standard predictive planner and 28% less than a predict-only one. No obstacle hits were seen. doi.org/10.1142/S23013…

@SpaceRhin0 @NASASpaceflight Great tracking on 137. Do you know if these McGregor runs are acceptance tests on new Raptor 3s, or re-tests on engines that already ran?

@westonmoyer @skyways @charlesacknin The $30,000 vs $1,000 per flight hour gap is the number that sells it. What payload does that $1,000 hour carry? Cost per kg per mile would make the helicopter comparison exact.

@pchees If all 150 upper stages a year fly, that is about 1.4 launches per pad per week across LC-36 and LC-36B. Which do you think limits cadence first: pad turnaround or upper stage engine supply?

@Drone_Wars_ 150 kg and 15 km one way is a clear CASEVAC card. Was that published at max weight, max range, or a mid trade for the combat trial?

@urbanairmobili1 ANH seats plus an LCR-110 IRS is the flight-critical supplier pass. Which cert gate do those two still need before Valo's next flight campaign?

Project Update. X70 turbine on our load cell. 69.8 N of thrust at 195,000 rpm.

Zipline started in October 2016 with blood deliveries to hospitals in Rwanda, where a late delivery has a real cost. Ten years of daily flights is what turns a drone demo into a service hospitals can plan around.

Zipline @zipline

2 days ago

Tomorrow we celebrate ten years of Zipline deliveries with a series of stories told by the people we serve.

1 9 77 58K 9

Sources: 1. rtx.com/news/news-cent… 2. shield.ai/shield-ai-and-… 3. news.lockheedmartin.com/2026-10-12-Sik… 4. breakingdefense.com/2026/10/coast-… 5. axios.com/2026/10/14/emb… 6. investors.beta.team/news-events/pr… 7. defensenews.com/industry/techw… 8. thehindu.com/news/cities/Hy…

Eight public numbers that moved aerospace and physical AI this week: 1. Raytheon: $729M LIDS C-UAS contracts for the U.S. Army, Qatar, and Jordan (first KuRFS and Coyote exports). 2. Shield AI X-BAT: Navy $150M OT mod plus a $150M company match. Public and private commitments now $400M. 3. Sikorsky Nomad 100 Fast Pack: 4 hybrid VTOL UAS in one 20-ft ISO. Two mission-ready in under 15 minutes. 4. U.S. Coast Guard first MQ-9B buy: 4 aircraft for $247.8M. Deliveries split across 2028 and 2029. Unarmed. 5. Eve Air Mobility: certification target December 2028. LOIs for up to 2,700 aircraft. 100 under contract with deposits. 6. BETA Utah eIPP: 5 days of ALIA CTOL radiopharma cargo demos. BETA says it is in 7 of 8 nationwide eIPP programs. 7. Anduril NGC2: $162.8M award (options to $1.8B over 5 years) to expand Lattice C2 to Army I Corps. 8. Drogo Aerospace JK 250e: DGCA type certification for its fixed-wing VTOL UAV (2-3 hour endurance). Six certified platforms.

Project Update. Landing fin v1 and v2 side by side. v2 gets a larger control surface, and the servo is now built into the fin. Fewer parts hanging outside means a cleaner fin and more control authority from the same footprint.

eVTOL battery cells are usually designed by build, test, repeat. Each loop is slow and costly. Avrion Battery Labs and Ionworks simulated 5,800 designs first and built 19 of them. The best cell for the high-power landing phase stored 21% more energy per kg than the baseline. It took two build loops instead of four. On a 97.3-minute mission test, the model's prediction was off by only 0.4 minutes. ionworks.com/blog/case-stud…