kctess5 / range_libc Goto Github PK

According to the paper and the pseudo code, the precomputation complexity should be O(|theta_discretization||width||height|) instead of O(|width|*|height|) for 2D grid.

When using CDDTCast, the max_range parameter is not always honored. The behavior seems to be limited to two scenarios.

Let's say max_range is set to 30.0:

1. An obstacle lies along the ray being cast at a true range of 50.0. CDDTCast::calc_range() will return 50.0 (incorrect, 30.0 expected).
2. No obstacle lies along the ray being cast (the map is open). CDDTCast::calc_range() will return 30.0 (correct).

It's not immediately obvious where the bug itself is because the algorithm is so complex and has lots of return statements. I'm not sure what a proper fix is but a workaround would be to clamp the max range returned by CDDTCast::calc_range().

The other raycasting methods Bresenham (BL), Ray Marching (RM), and Giant Lookup Table (GLT) are not affected.

Hi all!

I'm trying to build range_libc but got some problems related to CUDA_TOOLKIT (as shown below):

I carefully followed the building instructions so I don't know why I still got this problem. Is CUDA_TOOLKIT necessary for implementing Range_libc / Particle filter localization algorithm?
If not, how can I solve this problem?

@kctess5 @sertackaraman @arii
Thanks!

The method CDDT used to compute range is not consistent with the definition of ray casting as we only store LUT with rotation on obstacle around the origin of map.

Isn't that the exact method should be rotating obstacles around each grid cell?

HI dear author,
It's truly a honor to write a letter to you, I'm build your project nowadays and found the error when building as following, I wonder if there is a chance that you know the reason? :)

thank you
best regards to you
William

Shouldn't this be

self.thisptr = new OMap(<int>width,<int>height)

instead?

Otherwise at L145 we access grid[width-1] whereas grid is initialized to size height here

Hello Everyone,
I'm trying to use Range Libc package on NVIDIA Jetson Xavier, just like MIT Racecar configuration.
But after following your procedure to install it on Racecar Configuration, while installing pywrapper.
I'm getting this error,
/usr/local/cuda/bin/nvcc -I../ -I/usr/lib/python2.7/dist-packages/numpy/core/include -I/usr/local/cuda/include -I/usr/include/python2.7 -c ../includes/kernels.cu -o build/temp.linux-aarch64-2.7/../includes/kernels.o -arch=sm_20 --ptxas-options=-v -c --compiler-options '-fPIC' -w -std=c++11 -DUSE_CUDA=1 -DCHUNK_SIZE=262144 -DNUM_THREADS=256 nvcc fatal : Value 'sm_20' is not defined for option 'gpu-architecture' error: command '/usr/local/cuda/bin/nvcc' failed with exit status 1
Does anyone have any solutions to this issue?

Thank You,
Amey Hande

I'm interested with the range_libc. I want to use this library in the amcl package. What should I do?

Hi,

Thank you for develop this cool algorithm and open source it!
I found online CDDT and GPU CDDT available in the develop2 branch, are they usable?

Bests

Hi! First of all, great work!

I was wondering if you can provide an example in C++ of a particle filter using this library!

Cheers

I get the below error when running compile_with_cuda.sh on my Jetson TX2

nvcc fatal   : Value 'sm_20' is not defined for option 'gpu-architecture'
error: command '/usr/local/cuda/bin/nvcc' failed with exit status 1

Do you know this problem?