I hope i am posting this correctly. Maybe this can also be for sense_description as well. By default if I use the bigman.srdf and bigman_floating_base.urdf and any variant present in /config/bigman, I don't get correct tf from the mutisense head. Hence, when I start rotating the laser scanner with the scanner pointing on the ground, I get the all the ground points wrong, inclined at the same angle as the laser scanner. I did everything in the floating_base branch and all trials were in simulation.
I corrected this by removing all definitions of multisense except for /multisense/head from the the urdf mentioned in the config file which was bigman_floating_base.urdf. I then created another urdf put all the multisense components as a separate robot as you had once suggested. I also had to change the hokuyo_joint from a fixed to a continuous joint rotating about x-axis. If i did this joint change in the original advr urdf, i got seg-faults when running it. The I ran my own robot_state_publisher along with my code pointing to this new urdf and I got the laser scans correctly oriented. So can a similar correction be implemented in the repo or has this been already done??
liesrock
commented
6 years ago
@EnricoMingo
I hope i am posting this correctly. Maybe this can also be for sense_description as well. By default if I use the bigman.srdf and bigman_floating_base.urdf and any variant present in /config/bigman, I don't get correct tf from the mutisense head. Hence, when I start rotating the laser scanner with the scanner pointing on the ground, I get the all the ground points wrong, inclined at the same angle as the laser scanner. I did everything in the floating_base branch and all trials were in simulation.
I corrected this by removing all definitions of multisense except for /multisense/head from the the urdf mentioned in the config file which was bigman_floating_base.urdf. I then created another urdf put all the multisense components as a separate robot as you had once suggested. I also had to change the hokuyo_joint from a fixed to a continuous joint rotating about x-axis. If i did this joint change in the original advr urdf, i got seg-faults when running it. The I ran my own robot_state_publisher along with my code pointing to this new urdf and I got the laser scans correctly oriented. So can a similar correction be implemented in the repo or has this been already done??