code exited before exploration is finished.

[ning-zelin]

Hello,

I encountered an issue where the code exits unexpectedly before the exploration is completed. I suspect this might be due to the absence of frontier in the local area around the drone. However, there are still unexplored areas further away.

Thank you for your assistance. Looking forward to your reply.

Best regards

[haotongluo]

How to use Nvidia GPU in Docker containers? After the program runs, it cannot call the graphics card, resulting in very slow and crashing

[sherrypixel]

@haotongluo you probably need to adjust the "ENV LD_LIBRARY_PATH=/usr/lib/nvidia:$LD_LIBRARY_PATH" in the dockerfile to wherever 'libnvidia-ml.so' and 'libcuda.so' are located.

[sherrypixel]

@ning-zelin I am experiencing the same issue, were you able to find a reson why this is happening and how to fix it?

[haotongluo]

@ning-zelin我遇到了同样的问题，您能找到发生这种情况的原因以及如何解决它吗？

Thank you, I have already resolved the issue with the Nvidia GPU before, but I have also encountered the same situation where after the drone exploration, the message "/errt_node die" appears

[aakapatel]

Hi @sherrypixel and @haotongluo,

Thank you for your interest in our work. The E-RRT framework is a local planner and therefore does not have a global path planner strategy. We will provide a more suitable example environment where the local planner could be sufficient to demonstrate the continuous exploration. A fallback strategy would also be added that addresses local completion without seg fault.

Thanks, Akash

[sherrypixel]

@aakapatel thanks for your response. I was also wondering if we can use a camera like d435i instead of a Lidar?

[aakapatel]

Hi, @sherrypixel,

Yes as long as you have PointCloud2 input and a stable odometry, it should work. You may need to change the input topic for UFOMap. The computation for the information gain would be slightly off if you use the package as it is with camera point cloud. The current configuration computes information gain based on frustum field of view (usually for 3D LiDARs). If you want more accurate estimate, you can modify the information gain calculation to take into accound cone like FOV for the camera.

Thanks, Akash

[sherrypixel]

Thanks @aakapatel, I shall give this a shot.