Hi! SLAMBench is an excellent library.But I can not make it work well with AirSim datasets.My dataset contains rgb and depth images, as well as the groudtruth. I converted the dataset to TUM format and then make it into slambench file. Then I ran the following command in my terminal window:
./build/bin/pangolin_loader -i datasets/TUM/airsim/rgbd.slam -load ./build/lib/liborbslam2-original-library.so
ORBSLAM2 can run but did not output the correct results. And I find the same thing for the other SLAM algorithm.But these algorithms work well with TUM dataset downloaded from the Internet. So I think the problem is caused by my dataset. The screencap of pangolin is as follows:
Hi! SLAMBench is an excellent library.But I can not make it work well with AirSim datasets.My dataset contains rgb and depth images, as well as the groudtruth. I converted the dataset to TUM format and then make it into slambench file. Then I ran the following command in my terminal window: ./build/bin/pangolin_loader -i datasets/TUM/airsim/rgbd.slam -load ./build/lib/liborbslam2-original-library.so ORBSLAM2 can run but did not output the correct results. And I find the same thing for the other SLAM algorithm.But these algorithms work well with TUM dataset downloaded from the Internet. So I think the problem is caused by my dataset. The screencap of pangolin is as follows:
https://user-images.githubusercontent.com/59106182/147547814-649f4759-779d-4bec-a39d-0427320f220a.mp4
Can you give me some advice?