Make code changes to accommodate different chassis & leg dimensions

[nerdoug]

The existing code is based on the previous version of the chassis and legs, which had different dimensions. As well, changes to reduce the radial load on the hip servos will impact the software. Need to identify the areas that are impacted, and develop new code. Also need to standardize the symbols representing chassis dimensions, and where they're located in the source code.

[nerdoug]

Areas impacted:

• translations of local coords to global, and vice versa
• derivation of servo angles needed to achieve a specified toe position. This will be major, as there are several sets of simultaneous and quadratic equations involved. revevant routine is coordsToAngles() in include/flows.cpp
• maybe some flow processing

[nerdoug]

Survey of locations in source code that have references to physical robot dimensions:

configDetails.cpp

• servo calibration stuff in setupPerBotConfig

flows.h

• maxServoAngle
• toe safety box params safeMaxPosX...
• hexbot body measurements BT = 2.915 ...
• f_staticHomeX = 13.78 ...
• fp_frontToeHomeX= 3.8...
• sin_p45 = ... precalculations (need more for new chassis?)

flows.cpp

• setupFlows has f_dynGHomeX[1] = 18.62 ...
• and f_hip[1] = 8.87 ...
• mapDegToPWM has magic numbers in last line – servo type dependency
• anglesToCoords uses symbolic dimension references
• coordsToAngles has symbolic references
• globCoordsToLocal might have geometry dependencies? Refers to symbols. - fp_frontHipX
• localCoordsToGlobal refers to symbols: f_hipX[legnumber]
• handle_fo_newHomeLReset uses magic numbers to reset home position, and f_staticHomeX...

mqttBroker.cpp

• SSP command assumes servo PWM range = 110 – 490

Dumb question: are servos mounted differently, so rotation is in different direction for some?

• no, both old and new versions had shin at front of leg, so motors are acting the same.

As discussed in issue 17:

• robot physical dimension are defined in global_variables.cpp using macros for the actual values
• run time variables are initialized in a setup routine in configDetails.cpp

Issue 118, gets the dimension parameters defined and initialized in the right place. This issue, 83, addresses all the changes needed in the areas outlined above.

I'm about to add /physicalDesign/otherDesignInfo/dimensions-230209.ods/pdf which is a spreadsheet with the key physical measurements defined and recorded.

[nerdoug]

This has been implemented, but not tested. Basically re-used existing code with small changes:

• used new symbols starting with d$... for all physical dimensions, and used these symbols in place of previous symbols, or numeric measurements
• removed references to the displacement angle and distance that was part of the previous foot design where the toe was not on the vertical axis of the ankle servo. So we're using the same algorithms, except that the offset is zero degrees, and the toe is displaced by 0 cm from the vertical axis of the servo. It might be possible to simplify the code a bit, but in theory, "this should work"

• Changes for this uploaded to git in branch 118b

[nerdoug]

need to do testing to see if the algorithms still work before closing

[nerdoug]

noted that our existing movement scripts are no longer totally valid, although they made still work reasonably well. They specify most toe positions as absolute numbers, in either local or global coordinates, which may no longer apply. For example, many scripts start in the neutral stance, by specifying the toe position coordinates when the servos are all centered. However, because the dimensions of the robot have changed, going to the coordinates in the script will no longer get the neutral stance, and the servos will be slightly off center (knee and foot, anyway).

I don't think there's a way to do a program controlled conversion of the movement scripts, because it's hard to deduce what the original intention was.

Hmmm, one thing we need to do is update the movement script compiler. For example, it has a command something like "go to the home position" which now means a different set of coordinates. I'll try to put an issue in the right repository for that.

[nerdoug]

compiler issue for adjustments due to new dimensions is #52 in the compiler repository