Today I will try to detect some lanes..
Assumptions:
– We know the lane-width (plus minus)
– We are inside the middle of a lane
– We know the camera geometry
– Based on the turnrate of the IMU we can estimate the curvature of the street
– A line in pixels can be detected by a upward flank and a downward flank
Here are some exemplary results:
1) Of course, the best one first ;)
“
2) A little bit off, but still pretty good
3) If we change lanes, the prior assumption (we are inside of the lane) is violated, and it does not match any more (of course)
And as always, a results video:
Disussion
– Using only the turnrate as a prior for the curvature has a lot of flaws. The first is, it changes our priors too late, when we are already driving the curve, not if it is ahead already.
– Using the assumption that we are always in the middle of the lane clearly is not sufficient. The model should allow for lange-changes and „handing over“ lanes in the code with intermediate states.
– The tests that are done for the flanks in pixel-values should be more sophisticated ( currently they only check for a relative difference to detect high and low flanks)