polarisxq / scurm_sentrynavigation

SCURM火锅战队 24赛季哨兵导航

开源的初衷是想把自己的一些好的想法和大家一起分享，所以欢迎各位联系我和我讨论，欢迎PR，提issue🫠~

主要的创新：

1. 实现FAST-LIO2的重定位模式，不需要另外运行重定位算法。算力需求小，使得整个框架在NUC12上的资源占用在30%左右；精度高，定位误差小；内存占用不会随着时间增长。

2. 改进navigation2的故障恢复行为，卡住时机器人会向无碰撞的方向运动。

3. 实现底盘和坡面对齐(comming soon)

算法框架和思路详见技术报告

Docker镜像使用方法参阅DevcontainterGuide

包说明

Package Name	Description
✅auto_aim_interfaces
✅autonomous_exploration_development_environment	MODIFIED development env for cmu-series planner
✅BehaviourTree.CPP	MODIFIED BehaviourTree lib
✅cmd_chassis	- cmd_vel to chassis_cmd for communication and motion type of the chassis - exexute rotation command in chassis_link since the true value is unavailable
✅control_panel	a simple Qt GUI for simulating referee system
✅FAST_LIO	fastlio mapping
✅livox_ros_driver2	Driver for livox lidar
✅nav2_plugins - behavior_ext_plugins - costmap_intensity - velocity_smoother_ext	self defined nav2 plugins - an enhenced back_up action that move toward free space - 2 costmap_2d layer that use intensity filed of pointcloud msg rather than height (use with terrain analysis in autonomous_exploration_development_environment) - an enhenced velocity smoother that increase the speed on slope automatically (use with terrain analysis-pathNorm)
✅rm_decision_cpp	sentry desicion module based on BehaviourTree.CPP, implement nav2pose, topics2blackboard, anti_autoaim, attack etc. behaviours.

LAUNCH

MAPPING

• launch mapping node

ros2 launch sentry_bringup mapping.launch.py

• save map

# occupancy grid map
ros2 run nav2_map_server map_saver_cli -t /projected_map -f test_map --fmt png
# save pcd
ros2 service call /fast_lio_sam/save_map fast_lio_sam/srv/SaveMap
# or set the resolution by adding 
ros2 service call /fast_lio_sam/save_map fast_lio_sam/srv/SaveMap "{resolution: 0.0}"

• then terminate all nodes, pcd file will be saved in /PCD/scans.pcd

MAP PROCESSING

process pcd file

• drag the pcd file to CloudCompare(globalmap or scans.pcd), select the pointcloud in the left panel, then tools->clean->SOR filter, set the parameters (25,1 is a baseline) and apply
• select the processed pointcloud from last step, then tools->segmentation->Label Connected Components, set the parameters and apply
• pick out the CC#0(ususally this one), then tools->Other->Remove duplicate points, keep 1 point per 0.01-0.1m to reduce the size of the pointcloud
• select the processed pointcloud, then file->save as, select .pcd format

LAUNCH ALL

ros2 launch sentry_bringup bringup_all_in_one.launch.py