Indeed. Very same issue myself. Limiting to one turn sucks. On their forum I think I have seen a so called multi-turn fork on GitHub. Did not try it myself though.
I was discussing the very same topic with a friend of mine who claims to have implemented his own “vesc multi-turn firwware” too, precisely for a servo application (ie with gear reduction). Surprisingly his main design decision was because odrive was natively dual channel whereas vesc wasn’t + vesc was more compact. He offered to let me try it but said he had not decided whether or not he would opensource it
@p_v I do have some arduino code posted for the agv, but beware it’s really super ugly =>
Besides many things are missing: electronic schematics, hardware mechanical parts cad fiies, bill of material, photos, videos, documentation… with about 1500€ in parts it pulls 230kg trollies at about 0.4m/s (could go much faster but I have pedestrian traffic nearby). I also use a mode where it cruises with about 80kg load on the chassis.
How small is the delay you observe for full calibration?
If I remember right, there are indeed two ascii commands to issue but at least omy odrive machine, visually I did not notice the time difference. Besides, I use it on a corexy design (meaning you really have to sync both motors for nice sharp straight lines), and despite each motor consumes 1 command, i don’t use the trapezoidal planner (therefore bombard the odrive with POS commands using my own planner), and I have nice fast straight lines with significant torque too.
I think you should be just fine for a 2wd autonomous system with odrive (on paper at least).
What are: the mechanical reduction you plan + target cruise speed + motor model => you have selected?