Autonomy Sync Up - 9/15/2022
Meeting Attendants
Orson
Nico
Austin
Meeting Agenda
Discuss autonomy mission specs
Sync up on functionality, inputs and goals
Discuss potential disruption events
Brainstorm autonomy “plan of attack”
Main Challenges
Localization is incredibly important
System Inputs
GNSS coordinates (need to be accurate within 2m)
QR codes
Not really an input but continuing to track and publish GNSS coords and wheel speeds will help us detect stall events - may be worth looking into
System Outputs
LED Matrix Status
Gate Pass Status
System Functionality
Autonomy/Teleoperation Switches
Events to De-Risk
Motor Stall (rover gets stuck on a rock or something)
Camera gets dirty (object detection gets messed up)
Gimbal vibrational impact
is there someway we can detect heavy vibrations and camera shake
Autonomy Plan of Attack
Easy dub is to just turn on the zed2 and identify a QR code
Solve localization first. This is the most important part of autonomy. How can we actively track and publish GNSS coordinates
Look into differential GNSS
Look into ROS nav2 and localization
Get control and live video feed of the ZED2 camera
Ensure we can actually identify codes on our hardware (should be relatively easy)
Look into plans to identify QR codes
Some thoughts: drive to GNSS coordinates, sweep out to 20m, scan with gimbal to identify codes
Are there more efficient processes? What are the alternatives
Gate definition - what do we do after identifying a QR code?
gates are located 2m apart - what happens when one QR code for the gate is not within field of view, but what is?
determine if this is an issue through testing
we could drive to some calculated distance based off of FOV to always capture 2m horizontal viewing distance
How should we save the location of previous gates
later project - create active autonomy map that updates live during mission (what is the path of the rover)
Take a second look into what “difficult” terrain means. is this:
rocks that are traversable?
obstacles (probably not)
Action Items