Battleground? Are there many problems with Kalman filtering and preloaded maps of a path?

Not only that, GPS jamming can be overcome with better antennas (phased arrays of such, even retrofitted to existing commodity receivers) to eliminate jammers which are not more-or-less directly above the vehicle.

Excellent article on anti-jamming:

http://gpsworld.com/anti-jam-technology-demystifying-the-crp...

> CRPAs work by exploiting spatial diversity; that is, making use of the fact that the desired satellite signals, and the unwanted jamming signals, generally arrive from different directions. In simple terms, you create a spatial filter, one that removes signals that arrive from particular directions, whilst letting through signals from other directions. To achieve this, rather than use a single antenna, we use an array of antenna elements.

Well, yes. Maps aren't accurate (when it comes to moving foliage, as in the video, there maps are out of date within seconds, less in strong winds). Maps don't include many things that are moving, and maps just generally don't work very well at small scales (like the scale of an, even a large, UAV landing)

How about going low tech and just use gyroscopes to keep steady and fly in a straight line until it is unlikely that any jamming might occur?

The SR71 had a combination of star and inertial navigation. I imagine someone could build an astro-inertial navigation system today without much trouble.

Gyros drift pretty quickly (the cheaper the quicker). Yaw is hard to pin down when flying in a straight line even with a dual frequency gps.

Probably fine and less vulnerable in situations like simply trying to get a few km over a wall on flat ground. Far more complicated over terrain and trying to fly low.

Off-the-shelf open source consumer drone flight controllers already have enough sensors to achieve what you suggest.

I'm no expert but how autonomous works if there's no positioning methods involved? (To avoid GPS/GLONASS,... Jamming)

Fused dead reckoning can work, especially if you mix camera and IMU (or more than one IMU!).

Surprised nobodies mentioned celestial sensors.

Celestial navigation by imaging the sky, and combining that data with a clock and compass.

Very unreliable for ground based systems.

Cruise missiles have an optical terrain-following system, that would be a viable approach for drones.

They use radar.

https://en.m.wikipedia.org/wiki/TERCOM

Another technique not mentioned yet: compss augmented with local magnetic field features.

Inertial navigation and maps.

See the paper, but TLDR: stereo video depth sensing combined with fading previous measurements (so if you detected an obstacle at 1 o'clock, 2 meters out 1 second ago, it is now very likely at 2 o'clock, 1.3 meters out. This way you can limit yourself to just detecting things 2 meters out)