can't generate python code from symbolic Jacobian or about sympybotics HOT 1 CLOSED

Ok, this was simpler than I first thought. In fact the problem is the lack of documentation from my part :)

The function robot_code_to_func second argument must be a tuple of the type (list of intermediate variables, expression). The tuple second element is an expression (or a matrix of expression as it is in the case of the Jacobian), and the first element is a list of intermediate variables for which the expression depends on.

So, if we have no intermediate variables (this is the case), we must pass an empty list; something like ([], expression):

J_str = sympybotics.robotcodegen.robot_code_to_func('python', ([], rbt.kin.J[-1]), 'calc_jac', 'jacobian', rbtdef)

The tuple (list of intermediate variables, expression) has the same structure of the output of the Sympy cse (common subexpression elimination) function. This function detects common subexpressions in expressions and put them as auxiliary intermediate variables. This is something a C compiler will also do (when optimization is turned on) on the generated code, although the Sympy.cse can be slightly better as it has more mathematical knowledge about the expression.

For the dynamic model, the Sympybotics.RobotDynCode high level class uses an internal "iteractive" CSE I designed for the purpose. However, the geometric and kinematic parts has no CSE applied.
For the Jacobian, you can produce faster code by doing

J_code = sympy.cse(rbt.kin.J[-1])
J_str = sympybotics.robotcodegen.robot_code_to_func('python', J_code, 'calc_jac', 'jacobian', rbtdef)

(I have a personal script which generates the code for all the transformation matrices and all the Jacobians of all frames of a given robot as a single function, thus sharing the same common subexpressions in a very optimized way. I may release this code later.)

from sympybotics.