DDP, iLQG solvers return wrong-sized solution

[wxmerkt]

If the problem horizon is T, then:

• X is of size T
• U is of size T-1
• The solution in Solve in the solvers is resized to T without initialising or reinitialising to 0. At the end of solve, only the T-1th rows are getting set from the best U. That returns an additional row with a wrongly initialised memory.

Two possible solutions: a) The returned solution should be of size T-1 b) The final command should be a hold/inverse dynamics solution or zeros.

I tend to (a).

[traikodinev]

So the issue is this line here: https://github.com/ipab-slmc/exotica/blob/3462afd9928c02989754de09229f7ea6daebea23/exotations/solvers/exotica_ddp_solver/src/abstract_ddp_solver.cpp#L61

The solution is resized to T, not to T - 1, which is the issue (in the first paragraph you said T-1).

[traikodinev]

https://github.com/ipab-slmc/exotica/pull/656