Lessons Learned
Here are few points that the lab should address to stabilize our field deployments:
- Find a dedicated place for joysticks for each robot. We should not search for a joystick or be confused about which robot will be controlled
- Color code similar hardware
- Drill startup sequence
- Hard drive and data management
- Backup the NAS
- Automatic ROS bag record on the start of each mission
- Base station setup
- Drift at the beginning of the maps
- IMU and lidar must be fixed together and calibrated
- Many tests BEFORE the event
- Docker for each robots?
- Stop bringing personal laptop to field deployments
- If you deliver a robot, you need to control the full autonomy stack for tuning
- Packaging and tools
- Simple navigation behaviour
- Multiple person on the same module
- Synchronization of time, diagnostic
- Screen on robot for diagnostic
- Two bins for batteries: charged / uncharged
- A way to take full control of the robot for transportation (by pass safety limitations)
- Standard joystick button layout on all robots
- Colored hex keys (hallen keys)
- Special long pince
- Tie wrapped that can be removed
- Grinder that vibrates
- Product vs prototype
- Buck for power board
- Sensors used by CSIRO
- ANYmal
- Student va professional
- Ouster bad in smoke, high noise close range
- Lidar navigation was a key module for all teams