Open
Description
- Kinematic parameters for behaviors, controllers, etc
- Costmap plugin selection
- recovery speeds / sim times
- global localization leading the clock so TF always valid subtle item. TF tolerance tuning. why use TF over odom/pose topics (interpolating, current, etc).
- RPP tuning
- Goal checkers
- Progress checkers
- RL covariances in sensor and process (Tom?)
- smoother selection
- using dynamic parameters + gui to tune (tutorial?) (doisyg?)
- Costmap 2D (local size + speed, update rate, downsampling sensors, resolution, unknown space/inflate, obstacle/voxel params, etc)
- AMCL / weights (Fergs?)
- DWB (critics, generator, plugin, weights) (David?)
- Common critical performance mistakes (no static layer in local costmap for mislocalization/updates, not calibrating/tuning/filtering noisy sensors like depth cameras, tuning your low-level controller response on the platform to be good and not lag, etc)
- Caching obstacle heuristic (Smac Hybrid-A*, State Lattice)
- Controller selection
- Planner selection
- launch options (composition, headless, etc)
- Rotate in Place Behavior
- Robot Footprint vs Radius
- Inflation Potential Fields
- ... anything you think would be insightful!
Metadata
Metadata
Assignees
Labels
No labels