achang46
commented
2 years ago [Meeting, Alex-Arun, 4/15/22]
- Will re-structure such that gaits can be defined as longer state sequences (i.e. just long enough w.rt. 'practical' gaits/state sequences).
Per input from Caitlin:
- Max number of gaits we expect to use: 10
- Max state sequence length (we expect a gait to be defined by): 8
Began an interface description document that we can collaboratively finalize: https://docs.google.com/document/d/1NhcKvrKJ3Sy_6hmfkjQHOzd4s-_9TxEv1JlvnkAAXpk/edit#
"cycle_defn()" permits a gait to be defined as an arbitrary length state sequence (via serial input). This sequence is stored in the variable, 'cycle1'. But 'cycle1' is defined as a fixed length int8_t array (i.e. length 3) and, as result, doesn't appear to support storing arbitrary length sequences.
The Matlab demo script, API.m, actually sends a "define" (gait) command that defines a 4-state long sequence (even though 'cycle1' is only defined as a 3 length int8_t array). If API.m still ran fine and API_Msoro.ino never resulted in runtime errors (i.e. didn't guard against this assignment based on the variable definition), it seems the program could be writing that 4th state value (int8_t) to an unintended memory location.
Suggest restructuring such that gaits are stored in appropriately sized variables (i.e. able to support up to theoretical maximum state sequence length, given 4 binary actuators).