No the _ does not convey that meaning, however I often times name my private functions suchly.
You should indeed use defp

I also like the solution (keep forgetting about String.to_integer) however I still fail to see
how it passed all tests given that it does nothing concerning the tie breakers \o/

Thank you!

Ok, not an issue. Read more carefully the description:

minimal number of moves first, then minimal length of the path. You're doing the opposite.

(forget that, I'm tired. Reading again your post)

Thank you for the great cata!
Your reference solution in JavaScript does not always find all optimal solutions by steps.
Example:

  #    #  #     #    
    #        #       
      #  x     ##  ##
  E    #      #    # 
#    #  #   x#   #   
##   #   #x#     #  #
     #  # #   ## S   
           ##        
       x    #    #  #
      x            x 
    x  x x           
     #       # # # ##
   #   #   # x## x  #
#     # #        #   
# #      x   ###     
 x    #   x     #x # 
  #      # #  # #   
      #    #  # #    
##   # # #        # #

Possible optimal solutions (sorted by length):
dldrullul (length: 40)

Your solution: dldlllurdl (length: 30)

 # #     ##      #   #  
     #   #x  # # #      
##   # #  # #     #    #
#   # #                 
     #        # #       
              #x   # #  
x   #           ##      
 #           xx#  S  x#x
  #  #        ## # x  x 
  x      # #    ##   ##E
x           ## # # #  xx
            #        ## 
  # xx          x #  #x 
#     #          #  x   
              #    #  #x
  x         # x  x#  ## 
#     #   #     x#      

Possible optimal solutions (sorted by length):
urdd (length: 15)

Your solution: rdrrd (length: 7)

Is the _ for the _weight/1 to signal private use (like in python)?
If that's the case, you can use defp instead of def to declare it explicitely.

Nice work. I learnt the sirt hand $_weight/1 from this example
Thanks!