Code

Team CEBERUS has released a set of open-source packages, while more are on their way for release. Below find a selected list.

• Exploration Path Planning: https://github.com/ntnu-arl/gbplanner_ros
• Alternative exploration planner (only Phase I, II not in DARPA finals): https://github.com/ntnu-arl/mbplanner_ros 
• ANYmal Local Planner: https://github.com/leggedrobotics/art_planner
• Multi-robot Map alignment baseline: https://github.com/ethz-asl/maplab
• Elevation Mapping on GPU:  https://github.com/leggedrobotics/elevation_mapping_cupy
• Volumetric Mapping: https://github.com/ethz-asl/voxblox
• Visual-Inertial Odometry (part of CompSLAM - see comment below): https://github.com/ethz-asl/rovio
• Model Predictive Control for flying robots: https://github.com/ethz-asl/mav_control_rw

Note: The CERBERUS Simultaneous Localization And Mapping (SLAM) system, called CompSLAM, is currently not open-source. CompSLAM relies on the complementary fusion of a) LiDAR, b) Visual cameras, c) Thermal vision, d) IMU and e) when applicable leg odometry.

New packages will be released. Organization Github links:​

• https://github.com/ntnu-arl ​
• https://github.com/leggedrobotics ​
• https://github.com/ethz-asl​

---

Data

Team CERBERUS released the following datasets recorded during the DARPA SubT Challenge. Stay tuned for more to come!

• Final Event: Dataset & ​Meta-information.  Contains sensor data collected on-board the four ANYmal C robots used in the winning  final run.