Here's the problem: A small drone flies at you at 60mph / 100 km/h You have a camera, one edge processor, and must detect, classify, and track it at 30fps — against sky, trees, buildings. Starting from 100+ meters. If your pipeline drops a single frame at far range, the intercept fails. Sounds impossible? We think together we can solve this.
Now the second problem: Your guidance law needs to predict where that target will be in 0.5 seconds, compute the intercept geometry, and command the flight controller — with noisy IMU data, wind gusts, and a target that doesn't fly in straight lines. No GPS. No RF link. Just onboard compute.
We solved the guidance part in simulation. Our IBVS pipeline tracks and hits moving targets in SITL. The hardware prototype is built and ready. But the real engineering starts now.
We need:
* Computer Vision & Edge AI Engineer — YOLO on Jetson/Hailo/QRB5165. Detect, classify and track small targets at 100m+. Model pruning, TensorRT, sensor fusion. If you've pushed edge hardware to its absolute limits, this is your problem.
* GNC / Flight Systems Engineer — Image base visual servoing, proportional navigation, EKF/UKF state estimation, PX4 integration. Take what works in sim and make it survive real-world sensor noise.
* Simulation & Test Engineer — SITL/HITL, Monte Carlo campaigns, digital twins. 10,000 simulated engagements before anything leaves the ground.
* Technical Advisors — Drone warfare, GNC/aerospace, defense procurement. A few hours/month from the right person changes our trajectory.
We're a pre-seed defense startup in Munich building autonomous interceptor drones to stop FPV kamikaze drones. 2 founders. Salary + equity. No AI screening — we read every application.
We are building a UAV system to physically intercept fast-moving targets (100km/h+) using onboard compute only (Qualcomm QRB5165). We hit a wall regarding the Latency vs. Resolution trade-off and I’d love to hear some battle-tested opinions from the CV/Embedded community.The constraint: We need HD resolution to detect small targets at range, but running inference on full HD frames kills our...
We are building a UAV system to physically intercept fast-moving targets (100km/h+) using onboard compute only (Qualcomm QRB5165).We hit a wall regarding the Latency vs. Resolution trade-off and I’d love to hear some battle-tested opinions from the CV/Embedded community.The constraint: We need HD resolution to detect small targets at range, but running inference on full HD frames kills our...