Drone delivery at city scale, a deadline at the U.N. on autonomous weapons, and more police drone data. Here is what mattered in physical AI over the past day.
1. Zipline asks FAA for 130,000 drone deliveries a day over Arizona
An FAA draft would allow Zipline up to 65,000 drone deliveries a day in Phoenix and another 65,000 in Tucson. The agency’s notice lists two different deadlines for public comment.
Why it matters: A daily ceiling at this level points to drone delivery moving from limited pilots toward city-scale operations, and the comment period is the main opportunity for residents and local officials to weigh in.
2. U.N. talks on ‘killer robots’ run short on time
A U.N. conference on lethal autonomous weapons was running out of time on Friday after five days of talks on whether to begin negotiating regulations. Russia in particular slowed the discussions.
Why it matters: The outcome decides whether international rules for autonomous weapons move to formal negotiation, a question that bears directly on military uses of drones and robotics.
3. Watco expands Intramotev autonomous railcars to Texas
Short line railroad operator Watco will deploy Intramotev’s TugVolt autonomous freight railcar technology at a second site, in Seadrift, Texas. The agreement builds on the companies’ existing work at Watco’s terminal in Wood River, Illinois.
Why it matters: A repeat deployment by one of North America’s largest short line holding companies suggests autonomous rail is moving from a single site toward multi-site use.
4. DJI asks court to force an FCC ruling
DJI has asked the Ninth Circuit to compel the FCC to issue a ruling, according to Pilot Institute’s weekly update. The same roundup notes that BRINC now brings live drone feeds to iPhones and that SkyeBrowse has added a Realistic 3D feature.
Why it matters: DJI hardware is widely used by U.S. hobbyists, businesses and public-safety agencies, so a court-driven FCC decision could shape which drones those operators can buy and fly.
5. Nashville police drones beat officers to 98% of calls
Nashville police say their Skydio X10 drones reached 911 calls in about 2 minutes, compared with 15 minutes for officers. The trial ended after 30 flight days as overtime funding ran low.
Why it matters: Drone-as-first-responder programs are reporting results in several cities this week, and Nashville’s experience shows that staffing costs, not only hardware, can limit them.
6. AI agent links experiment planning with robotic control
Researchers describe an adaptive AI agent that connects experimental planning with robotic control, tested on a wastewater treatment task. The work targets a gap: most AI research agents are disembodied and depend on humans or separate automated systems to carry out experiments.
Why it matters: Closing the loop between planning and physical lab work is a central goal for automated, self-driving laboratories.
7. UCLA police cleared for five Skydio drones for LA28 village
The UC Regents approved growing UCLA’s police drone fleet from three Skydio X10s to five to cover the LA28 Athletes Village. The first three drones cost $216,543 over ten years.
Why it matters: The 2028 Olympics are shaping up as a high-profile test of police drone use around a major event. See upcoming industry dates on our events page.
Quick hits
- New Castle County, Delaware police drones arrived first at 756 of 844 calls over six months and 1,214 missions; the policy sets no video retention period.
- St. Louis police drones reached the scene first on 18 of 58 flights since Nov. 4.
- Amazon’s Prime Air pad in Jacksonville seeks to cut a 500-foot helicopter setback to 29 feet.
- Pinehurst, N.C. approved a free year of BRINC police drones, with a quote billing $159,998 a year for five more years unless the village gives notice.
- Video Friday: a quadrotor folds from 31.3 cm to 14.1 cm wide on contact to fly through a 15 cm opening.
Get the Daily and Weekly Brief by email — subscribe free. Every item links to our summary and the original reporting. Brief compiled by Physical AI News.
Newsletter
Get the Physical AI brief
The day’s most important robotics and embodied-AI stories, in your inbox.


