Michael Kösel

Using SoA instead of AoS particles reduced the runtime by around 50ms. Changing the grid cells to SoA gives a small improvement. See https://github.com/TheCodez/dynamic-occupancy-grid-map/tree/soa Big improvements coming soon :smile:

Might take some time I'm thinking of using this abstraction https://asc.ziti.uni-heidelberg.de/sites/default/files/research/papers/public/St11ASX_CUDA.pdf to keep the AoS syntax. But at least with the current implementation I know that SoA is the way...

Mostly blocked by https://github.com/TheCodez/dynamic-occupancy-grid-map/issues/70 because of the PCL dependency.

Hi @synsin0, currently the ROS branch is unfishished and highly outdated. I don't know when I can finish it, as I have very little time at the moment. Your best...

@ShepelIlya Thank you. I will take a look as soon as possible.

@idlebear have you seen a noticable runtime performance difference with SoA grid cells? @ShepelIlya no worries, I will include a memory leak fix once I have some more time to...

@idlebear I'm seeing no improvement either, which is unfortunate as I was hoping that this would get the performance closer to the paper. As for the the trails and noise,...

Can you try installing clang-tidy using `apt install clang-tidy`?

Alternatively, if you don't want to install clang-tidy you can do the following: ``` mkdir -p build && cd build cmake -DCMAKE_BUILD_TYPE=Debug ../dogm make ```

To me it seems as if your compiler is not correctly set