Todo List

Class BenchmarkCFG

the getter functions do not check if the types are admissible.

does not have support for multiple starts or goals.

the way ctor parameters are given could be improved (as declareParams in OMPL - command pattern).

Class Cell

No checks are done (out of bounds, etc) to improve efficienty. Overload functions to add optional input checking.

Class console

a member method like string but with const char* is missing.

Class DDQM< grid_t >

implement a more robust goal point stopping criterion.

Member EikonalSolver< grid_t >::solveEikonal (const int &idx)

given that this sorts a small vector, a n^2 methods could be better. Test it.

Class FM2< grid_t, heap_t >

Include support to other solvers (GMM, FIM, UFMM). It requires a better way of setting parameters.

Class FM2Star< grid_t, heap_t >

Include support to other solvers (GMM, FIM, UFMM). Requires theoretical work on heuristics on these methods.

Class FMCell

Overload functions to add the option of input checking. No checks are faster.

Member FMM< grid_t, heap_t >::FMM (HeurStrategy h=NOHEUR)

automate the naming depending on the heap.

Member FMMStar< grid_t, heap_t >::FMMStar (HeurStrategy h=TIME)

automate the naming depending on the heap.

Class FMUntidyQueue< cell_t >

save buckets here as a hash table instead of saving them in FMCell.

Class FSM< grid_t >

implement a more robust goal point stopping criterion.

Class GridPlotter

include checks which ensure that the grids are adecuate for the functions used.

generalize plotMapPaths to work with multiple (n>2) paths.

Class LSM< grid_t >

implement a more robust goal point stopping criterion.

Class MapLoader

Include checks which ensure that the grids are adequate for the functions used.

When loading an empty grid, cells are not correctly initialized. Although it works properly, values are not set to Inf (that is why we have Nan values in obstacles.)

Class nDGridMap< T, ndims >

Neighbors precomputation could speed things up.

Improve coord2idx function in order to just pass n coordinates and not an array.

Create d_ with 1 and d_[1] size of X, d_[2] size of Y, etc, to generalize dimensions.

Class Solver< grid_t >

Init and goal points are not checked to be in the map.