A relatively simple way is to first divide the determined range into n cells, and then get a point in each interval with a random number, which will be relatively sparse. Adding a padding to the interval can prevent two points from being too close to each other.

Random must have repetition, and the so-called "non-repetitive random" is actually a shuffle.

the idea I can think of is that if you need to get N sets of longitudes and latitudes, first divide the circle 1 km from the current longitude and latitude [x, y] into N parts, and then find out the central points of the N parts. and then random on these N central points.

I hope I can help you.

was in the form of random number in the wrong way. Later, it was adopted to obtain the maximum and minimum longitude and latitude first, then take the decimal part of each maximum and minimum longitude and latitude, enlarge N times random, and then reduce N times to get a set of random longitude and latitude