Multiagent CPP API

src/ale/ale_interface.cpp

@@ -243,7 +243,7 @@ bool ALEInterface::game_over(bool with_truncation) const {
   return with_truncation ? environment->isTerminal() : environment->isGameTerminal();
 }
 
-// Indicateds if the episode has been truncated.
+// Indicates if the episode has been truncated.
 bool ALEInterface::game_truncated() const { return environment->isGameTruncated(); }
 
 // The remaining number of lives.
@@ -259,9 +259,13 @@ int ALEInterface::lives() {
 // user's responsibility to check if the game has ended and reset
 // when necessary - this method will keep pressing buttons on the
 // game over screen.
-// Intentionally set player B actions to 0 since we are in single player mode
-reward_t ALEInterface::act(Action action, float paddle_strength) {
-  return environment->act(action, PLAYER_B_NOOP, paddle_strength, 0.0);
+reward_t ALEInterface::act(
+  Action a_action,
+  float a_paddle_strength,
+  Action b_action,
+  float b_paddle_strength
+) {
+  return environment->act(a_action, b_action, a_paddle_strength, b_paddle_strength);
 }
 
 // Returns the vector of modes available for the current game.

src/ale/ale_interface.hpp

@@ -86,7 +86,12 @@ class ALEInterface {
   // user's responsibility to check if the game has ended and reset
   // when necessary - this method will keep pressing buttons on the
   // game over screen.
-  reward_t act(Action action, float paddle_strength = 1.0);
+  reward_t act(
+    Action a_action,
+    float a_paddle_strength = 1.0,
+    Action b_action = NOOP,
+    float b_paddle_strength = 0.0
+  );
 
   // Indicates if the game has ended.
   bool game_over(bool with_truncation = true) const;

src/ale/common/Constants.cpp

@@ -22,51 +22,33 @@ namespace ale {
 
 std::string action_to_string(Action a) {
   static std::string tmp_action_to_string[] = {
-      "PLAYER_A_NOOP",
-      "PLAYER_A_FIRE",
-      "PLAYER_A_UP",
-      "PLAYER_A_RIGHT",
-      "PLAYER_A_LEFT",
-      "PLAYER_A_DOWN",
-      "PLAYER_A_UPRIGHT",
-      "PLAYER_A_UPLEFT",
-      "PLAYER_A_DOWNRIGHT",
-      "PLAYER_A_DOWNLEFT",
-      "PLAYER_A_UPFIRE",
-      "PLAYER_A_RIGHTFIRE",
-      "PLAYER_A_LEFTFIRE",
-      "PLAYER_A_DOWNFIRE",
-      "PLAYER_A_UPRIGHTFIRE",
-      "PLAYER_A_UPLEFTFIRE",
-      "PLAYER_A_DOWNRIGHTFIRE",
-      "PLAYER_A_DOWNLEFTFIRE",
-      "PLAYER_B_NOOP",
-      "PLAYER_B_FIRE",
-      "PLAYER_B_UP",
-      "PLAYER_B_RIGHT",
-      "PLAYER_B_LEFT",
-      "PLAYER_B_DOWN",
-      "PLAYER_B_UPRIGHT",
-      "PLAYER_B_UPLEFT",
-      "PLAYER_B_DOWNRIGHT",
-      "PLAYER_B_DOWNLEFT",
-      "PLAYER_B_UPFIRE",
-      "PLAYER_B_RIGHTFIRE",
-      "PLAYER_B_LEFTFIRE",
-      "PLAYER_B_DOWNFIRE",
-      "PLAYER_B_UPRIGHTFIRE",
-      "PLAYER_B_UPLEFTFIRE",
-      "PLAYER_B_DOWNRIGHTFIRE",
-      "PLAYER_B_DOWNLEFTFIRE",
-      "__invalid__",  // 36
-      "__invalid__",  // 37
-      "__invalid__",  // 38
-      "__invalid__",  // 39
-      "RESET",        // 40
-      "UNDEFINED",    // 41
-      "RANDOM",       // 42
+      "NOOP",
+      "FIRE",
+      "UP",
+      "RIGHT",
+      "LEFT",
+      "DOWN",
+      "UPRIGHT",
+      "UPLEFT",
+      "DOWNRIGHT",
+      "DOWNLEFT",
+      "UPFIRE",
+      "RIGHTFIRE",
+      "LEFTFIRE",
+      "DOWNFIRE",
+      "UPRIGHTFIRE",
+      "UPLEFTFIRE",
+      "DOWNRIGHTFIRE",
+      "DOWNLEFTFIRE",
+      "RESET",        // 18
+      "UNDEFINED",    // 19
+      "RANDOM",       // 20
+      "__invalid__",  // 21
+      "__invalid__",  // 22
+      "__invalid__",  // 23
+      "__invalid__",  // 24
   };
-  assert(a >= 0 && a <= 42);
+  assert(a >= 0 && a <= 24);
   return tmp_action_to_string[a];
 }
 

src/ale/common/Constants.h

@@ -24,53 +24,34 @@ namespace ale {
 
 // Define actions
 enum Action {
-  PLAYER_A_NOOP          = 0,
-  PLAYER_A_FIRE          = 1,
-  PLAYER_A_UP            = 2,
-  PLAYER_A_RIGHT         = 3,
-  PLAYER_A_LEFT          = 4,
-  PLAYER_A_DOWN          = 5,
-  PLAYER_A_UPRIGHT       = 6,
-  PLAYER_A_UPLEFT        = 7,
-  PLAYER_A_DOWNRIGHT     = 8,
-  PLAYER_A_DOWNLEFT      = 9,
-  PLAYER_A_UPFIRE        = 10,
-  PLAYER_A_RIGHTFIRE     = 11,
-  PLAYER_A_LEFTFIRE      = 12,
-  PLAYER_A_DOWNFIRE      = 13,
-  PLAYER_A_UPRIGHTFIRE   = 14,
-  PLAYER_A_UPLEFTFIRE    = 15,
-  PLAYER_A_DOWNRIGHTFIRE = 16,
-  PLAYER_A_DOWNLEFTFIRE  = 17,
-  PLAYER_B_NOOP          = 18,
-  PLAYER_B_FIRE          = 19,
-  PLAYER_B_UP            = 20,
-  PLAYER_B_RIGHT         = 21,
-  PLAYER_B_LEFT          = 22,
-  PLAYER_B_DOWN          = 23,
-  PLAYER_B_UPRIGHT       = 24,
-  PLAYER_B_UPLEFT        = 25,
-  PLAYER_B_DOWNRIGHT     = 26,
-  PLAYER_B_DOWNLEFT      = 27,
-  PLAYER_B_UPFIRE        = 28,
-  PLAYER_B_RIGHTFIRE     = 29,
-  PLAYER_B_LEFTFIRE      = 30,
-  PLAYER_B_DOWNFIRE      = 31,
-  PLAYER_B_UPRIGHTFIRE   = 32,
-  PLAYER_B_UPLEFTFIRE    = 33,
-  PLAYER_B_DOWNRIGHTFIRE = 34,
-  PLAYER_B_DOWNLEFTFIRE  = 35,
-  RESET                  = 40, // MGB: Use SYSTEM_RESET to reset the environment.
-  UNDEFINED              = 41,
-  RANDOM                 = 42,
-  SAVE_STATE             = 43,
-  LOAD_STATE             = 44,
-  SYSTEM_RESET           = 45,
-  LAST_ACTION_INDEX      = 50
+  NOOP                   = 0,
+  FIRE                   = 1,
+  UP                     = 2,
+  RIGHT                  = 3,
+  LEFT                   = 4,
+  DOWN                   = 5,
+  UPRIGHT                = 6,
+  UPLEFT                 = 7,
+  DOWNRIGHT              = 8,
+  DOWNLEFT               = 9,
+  UPFIRE                 = 10,
+  RIGHTFIRE              = 11,
+  LEFTFIRE               = 12,
+  DOWNFIRE               = 13,
+  UPRIGHTFIRE            = 14,
+  UPLEFTFIRE             = 15,
+  DOWNRIGHTFIRE          = 16,
+  DOWNLEFTFIRE           = 17,
+  RESET                  = 18, // MGB: Use SYSTEM_RESET to reset the environment.
+  UNDEFINED              = 19,
+  RANDOM                 = 20,
+  SAVE_STATE             = 21,
+  LOAD_STATE             = 22,
+  SYSTEM_RESET           = 23,
+  LAST_ACTION_INDEX      = 24
 };
 
-#define PLAYER_A_MAX (18)
-#define PLAYER_B_MAX (36)
+#define ACTION_MAX (18)
 
 std::string action_to_string(Action a);