This repostory contains software for the virtual track of the DARPA SubT Challenge. Within this repository you will find Gazebo simulation assets, ROS interfaces, support scripts and plugins, and documentation needed to compete in the SubT Virtual Challenge.
Other
309
stars
97
forks
source link
Model submission for Husky sensor config 3, 4 by CTU-CRAS-Norlab. #1047
Sensor configs 3, 4 are the CTU-CRAS build used in Finals. Sensor configs 1, 2 (already submitted and accepted) are the Norlab build that was a backup for the Finals and was not used in the end.
The motion characteristics should remain the same as sensor config 1 (just total mass is lower by 3 kg). You can check that comparing the relevant parts of model.sdf. I also fixed a weird inertia of the base link.
The sensors are the same used on MARV, Absolem 6 and other robots.
However, I did some tweaks to the Ouster model:
I decreased horizontal resolution to 1024 points/revolution as the 2048 mode is impossible to cool on the real robot and the lidar goes to shutdown very quickly.
I increased min range of the lidar to 25 cm. This corresponds to reality (they advertise 0 cm min range, but that is a lie/promise of a not yet released feature). This also prevents the lidar from outputting the weird artifacts observed e.g. in https://github.com/osrf/subt/issues/888#issuecomment-853357896 . I'll investigate if this isn't the simple way to fix Ousters on all of our robots without needing to dig into OGRE or changing lidar geometry assets.
I manually checked the look of collisions and inertia and they look okay. The inertia of sensor config 1's base link was weirdly rotated, so I fixed it in sensor configs 3,4. But maybe it is something to check for all the other Husky models. Cameras look okay. Lidar looks okay. Breadcrumbs are not present in config 3, and are usable with config 4. They fall from a place close to the dropper (it is intentionally in front of the robot). Battery drains. ROS-Ign bridge was checked for sensors, driving and breadcrumbs. Driving behaves as expected.
peci1
commented
2 years ago
Sensor configs 3, 4 are the CTU-CRAS build used in Finals. Sensor configs 1, 2 (already submitted and accepted) are the Norlab build that was a backup for the Finals and was not used in the end.
The motion characteristics should remain the same as sensor config 1 (just total mass is lower by 3 kg). You can check that comparing the relevant parts of model.sdf. I also fixed a weird inertia of the base link.
The sensors are the same used on MARV, Absolem 6 and other robots.
However, I did some tweaks to the Ouster model:
I manually checked the look of collisions and inertia and they look okay. The inertia of sensor config 1's base link was weirdly rotated, so I fixed it in sensor configs 3,4. But maybe it is something to check for all the other Husky models. Cameras look okay. Lidar looks okay. Breadcrumbs are not present in config 3, and are usable with config 4. They fall from a place close to the dropper (it is intentionally in front of the robot). Battery drains. ROS-Ign bridge was checked for sensors, driving and breadcrumbs. Driving behaves as expected.