Day 3: Lifeline
Welcome to Day 3. Today drones save lives.
In Rwanda, a Zipline drone is in the air every 4 minutes. It's delivering blood, antivenom, or vaccines to a clinic that doesn't have a road. In California, REACH Air Medical helicopters cost $30,000 a flight and take 45 minutes to reach a Trinity Alps trailhead. A drone could deliver a tourniquet in 12.
You'll use AI tools to plan a real-feeling mission, then fly it. Meanwhile half the class drives a ground robot through a post-fire structure assessment — the actual job Cal Fire ground teams now do before sending crews into a burned building.
Two real numbers to start the day:
- Zipline has delivered 1.4 million packages in Rwanda, Ghana, Nigeria, and the United States — mostly blood, vaccines, and medicines. Their drone is the size of a small carry-on suitcase. It cruises at 60 mph, flies for an hour, and drops the package by parachute within a 6-foot circle of the target.
- REACH Air Medical Services, based at Redding Airport, flies helicopters all over the North State. The closest medevac to a hiker on Lassen Peak is REACH 5, based in Susanville. Best case from launch to patient: 30+ minutes. A drone with a tourniquet, an EpiPen, or naloxone could be there in 8.
The technology is real. The companies are real. The jobs are real. Today you'll design and fly the junior version of a mission like this — with AI doing the heavy planning work alongside you.
Zipline is the largest autonomous delivery company on Earth. They hire flight operations specialists (~$70K starting), mission planners ($90K), and software engineers ($150K+). Many start as flight ops technicians with just a Part 107 and some drone experience — the kind of resume you'd have by the end of next summer if you wanted it.
Before you use AI for anything important, you need to know what it's actually good at — and what it isn't.
AI is great at: brainstorming, summarizing, drafting plans, translating ideas to step-by-step instructions, generating images. AI is bad at: real-time data, exact math, knowing what's happening today, knowing your specific drone. A good operator knows the difference and uses it for the right job.
This morning, you're going to use AI to plan the mission you fly in an hour. The plan AI gives you will be a starting draft. You're the pilot — your job is to read it critically, fix what's wrong, and make it real.
On your phone, pick one AI to use today. They're all free. Pick the one you already have an account for, or the one that looks easiest to log into right now.
Stuck without an account? Use Copilot — it works without a sign-in.
Type this into your AI exactly as written:
Read the answer. Ask yourself:
- Is it actually 4 sentences? (AIs often miss simple constraints.)
- Did it answer the second part — what makes mission planning hard?
- Is anything in it obviously wrong?
If you weren't satisfied, type: Try again — exactly 4 sentences this time.
This is prompt engineering. The skill of getting useful work out of AI isn't typing once and accepting whatever comes back. It's having a real conversation — push back, narrow the constraints, ask for revisions. People are getting paid $150K to do exactly this.
Now the real work. Copy this prompt into your AI and adjust the bracketed parts for your team:
- Send the prompt.
- Read what it gives you. Don't accept it. Pick one thing that's wrong, vague, or unrealistic.
- Reply:
Rewrite step 2 — the flight path is too vague. Tell me exact distances and turns I could write into block code. - Keep iterating until you have something you'd actually fly.
- Screenshot the final plan. You'll use it at the Tello station in 30 minutes.
AI will sometimes say a Tello can fly miles, lift kilograms, or drop with GPS precision. None of that is true. When AI confidently invents specs, that's called hallucinating. Spot it. Push back. The pilot is still you.
Break + station move. Same groups as Day 2. Group A starts at Tello LIFELINE Drop. Group B starts at RoboMaster Post-Fire. Swap at 11:45.
10:40 – 10:50The "trailhead landing zone" is taped on the gym floor — a 1-meter square. The "patient" is 6 meters away on the other side of the gym. Your job: fly the route from start to the LZ, drop your payload inside the square, and return safely.
You will fly with a payload taped to the top of your drone — a small paper card representing the medical package. When your drone executes a programmed flip at the LZ, the card falls off. That's your delivery.
- Your Tello + 2 batteries + DroneBlocks-loaded phone
- 📋 LIFELINE Mission Log (1 per pair)
- 5 "medical payload" cards (small index cards, lightly taped on top)
- A 1-meter LZ taped on the gym floor at the target
- Your AI mission plan screenshot (from the lab earlier)
Open your AI mission plan from earlier. Open DroneBlocks. Translate the plan into blocks.
Your mission should include, in order:
Why the flip drops the card: the propwash + sudden rotation knocks the lightly-taped card off the top. It's not how real Zipline works (theirs uses a parachute release), but it's the cleanest no-extra-hardware solution for a Tello.
Read what AI told you. Did it give you distances? Are those distances even possible for a Tello inside the gym? If AI said "fly 50 meters" — the gym isn't 50 meters. Cut it.
- First run: no payload card. Just verify your path lands in the right place.
- Watch where the drone ends up after the flip. If it's drifting off the LZ, your distance is wrong.
- Adjust the forward/back distance until your drone is hovering over the LZ before the flip.
Tape a payload card flat on top of your Tello, between the front-camera bump and the rear. Use one small piece of masking tape — not a wrap. The card should sit, not be glued.
The scoring:
- Payload lands inside the LZ square = full delivery
- Payload lands within 1 foot of the LZ = partial delivery (Zipline counts this as success)
- Payload doesn't release = retry — tape was too tight
- Drone crashes = mission abort, log the failure
Each pair runs at least 3 delivery attempts. Mark each on your Mission Log.
Take a video of your best delivery. Upload to Padlet: [PADLET LINK], "Day 3 — LIFELINE Drops" column. Caption with your team name, payload type (antivenom, tourniquet, etc.), and your distance from LZ.
Rich pulls one group of four outside during each rotation. Day 3 outdoor mission: a longer-range "scouting flight" over the football field — what a Cal Fire UAS team would do to find the edge of a fire perimeter.
After a structure fire, the fire is out — but the building isn't safe. Floors may be burned through. Walls may collapse. There can be hot spots smoldering inside walls that re-ignite hours later. Carbon monoxide, propane leaks, downed wires.
Cal Fire and increasingly local FDs send a ground robot in first. The robot has a thermal camera and a regular camera. It rolls through the structure, finds hot spots, identifies hazards, and the team uses that map to decide if (and where) it's safe to send humans.
Today you do the no-thermal-camera version: drive the RoboMaster through a "burned residence" obstacle course, find the hidden hot-spot markers, and document each one.
Cal Fire Heavy Fire Equipment Operator and Structure Damage Assessment teams use ground robots after every major structure fire in the state. The robots came in after the Camp Fire (Paradise, 2018) when assessment crews needed to enter 18,000+ destroyed homes. The robot operator is usually a fire technology graduate from Shasta College or Butte College — local 2-year degree, leads directly to a state job.
- 1 RoboMaster S1 (2–3 students per robot)
- Program iPad with RoboMaster app
- "Burned residence" obstacle course — chairs/boxes/mats representing collapsed structure
- 5 "hot spot" markers hidden in the course (printed orange triangles with location codes)
- 📋 Damage Assessment Log (1 per team)
You drove RoboMasters yesterday for SAR. Today is faster: you already know the controls.
Before you drive, look at the course from the outside for 2 minutes. Sketch a rough map on the back of your Damage Assessment Log. Mark where you think you'll go first.
Why plan first: a real structure damage assessment team has aerial photos and floor plans before they roll in. Plan before drive. Pilots who skip planning crash robots.
Same rules as Day 2 SAR: driver uses camera only. No looking at the robot. Spotter calls obstacles. Reader logs hot spots.
For each hot spot you find:
- Position the camera so the orange marker is centered in the iPad screen.
- Screenshot the iPad. That's your "thermal capture" image.
- Log the marker's code on your Damage Assessment Log + describe what's around it ("under collapsed beam," "behind appliance").
- Rotate driver every 7 minutes. Everyone drives.
Goal: find all 5 hot spots in the time available. Bonus credit if you do it with a planned route (not random wandering).
At a real scene, the robot operator briefs the incident commander when the sweep is done. You're going to do the same — in writing.
On your Damage Assessment Log, write a 3-sentence briefing answering:
- How many hot spots did you find, and where (general areas of the structure)?
- Which area is the most dangerous for ground crews to enter, and why?
- What would you want a real thermal camera for that the regular camera couldn't tell you?
Upload the briefing + your sketch + one hot spot photo to Padlet: [PADLET LINK], "Day 3 — Damage Assessment" column.
Swap stations. Same drill as Day 2 — leave equipment, move yourselves.
11:40 – 11:45Same station instructions above — just at the swapped station. Phantom rotation continues.
Everyone back to the center of the gym. Phones out — we're watching the Padlet on the projector.
Each team picks one deliverable to share:
- Your best LIFELINE drop video OR
- Your damage assessment briefing + best hot-spot photo
We pull yours up on the projector. 90 seconds per team. Answer:
- What was the mission, in one sentence?
- What did AI help you with, if anything?
- What surprised you?
Drones down. iPads in. Stand by your station.
- Designed a humanitarian mission with an AI co-pilot — and pushed back when AI got it wrong.
- Flew an autonomous delivery — same shape mission as a $90K Zipline pilot.
- Briefed a chief on a structure assessment — same skill Shasta College fire tech grads sell to Cal Fire on day one.
Day 4 you become a founder. You'll build a North State drone services company — name it, brand it (AI image gen comes back), figure out what you sell, who pays for it, and what it takes to start. Then you pitch it to the room. The Elevate Scholarship → Part 107 → small business pathway is real, and Day 4 is when we walk you through it.
Last day tomorrow. Bring your TRUST cert and your best ideas.