Result without ROS and trajectory are different

[Nabetalion]

I try to use SVO without ROS. I follow the README.md and run the test_pipeline.cpp, but result are different.

I used the dataset: sin2_tex2_h1_v8_d.tar.gz. And, I compute it and got a result. The results in frame 185 are: Trajectry: 3.680000 frame_000185 24.6517 2.8944 2.0000 0.8648 -0.0374 0.0000 0.0000 Computed result: 11.6264 4.94662 -0.028772 I guess we should obtain similar value between 24.6517 and 11.6264. The difference is double size.

I add the following code shown below in "test_pipeline.cpp". This code is made from the "https://github.com/uzh-rpg/rpg_svo/wiki/Notation" // access the pose of the camera via vo_->lastFrame()->T_f_w_. Eigen::Vector3d p; p = vo_->lastFrame()->T_f_w_.inverse().translation(); std::cout << "pose:" << p[0] <<"\t"<< p[1]<<"\t"<<p[2]<<std::endl;

My questions are:

• This result is true?
• How to obtain the position, posture and seed of map without ROS?
• Can we use other dataset for SVO without ROS?

[jixiedaima]

I can get get trajectory information,but I can't use it to evaluate the result. Do you know why?

[Sho412]

I have same issues.

base file: test_pipeline.cpp insert position: after vo_addImage(img,0.01*img_id); insert code: Vector3d p; p=Vector3d(vo_->lastFrame()->T_f_w_.inverse().translation();

Computed result: 11.55644 4.921084 -0.008531

Is there any solution?

[Sho412]

I'd found the result without ros is same as the result which is mention in other issue. https://github.com/uzh-rpg/rpg_svo/issues/125

Need to -Set the initial position to the one of the ground-truth trajectory. -Set the scale. or -Alignment after calculation