[🐛 🔨 ] set_action_for_agent expects a ActionTuple with batch size 1.

ml-agents-envs/mlagents_envs/base_env.py

@@ -410,28 +410,20 @@ def random_action(self, n_agents: int) -> ActionTuple:
         return ActionTuple(continuous=_continuous, discrete=_discrete)
 
     def _validate_action(
-        self, actions: ActionTuple, n_agents: Optional[int], name: str
+        self, actions: ActionTuple, n_agents: int, name: str
     ) -> ActionTuple:
         """
         Validates that action has the correct action dim
         for the correct number of agents and ensures the type.
         """
-        _expected_shape = (
-            (n_agents, self.continuous_size)
-            if n_agents is not None
-            else (self.continuous_size,)
-        )
+        _expected_shape = (n_agents, self.continuous_size)
         if actions.continuous.shape != _expected_shape:
             raise UnityActionException(
                 f"The behavior {name} needs a continuous input of dimension "
                 f"{_expected_shape} for (<number of agents>, <action size>) but "
                 f"received input of dimension {actions.continuous.shape}"
             )
-        _expected_shape = (
-            (n_agents, self.discrete_size)
-            if n_agents is not None
-            else (self.discrete_size,)
-        )
+        _expected_shape = (n_agents, self.discrete_size)
         if actions.discrete.shape != _expected_shape:
             raise UnityActionException(
                 f"The behavior {name} needs a discrete input of dimension "

ml-agents-envs/mlagents_envs/environment.py

@@ -365,9 +365,9 @@ def set_action_for_agent(
         if behavior_name not in self._env_state:
             return
         action_spec = self._env_specs[behavior_name].action_spec
-        num_agents = len(self._env_state[behavior_name][0])
-        action = action_spec._validate_action(action, None, behavior_name)
+        action = action_spec._validate_action(action, 1, behavior_name)
         if behavior_name not in self._env_actions:
+            num_agents = len(self._env_state[behavior_name][0])
             self._env_actions[behavior_name] = action_spec.empty_action(num_agents)
         try:
             index = np.where(self._env_state[behavior_name][0].agent_id == agent_id)[0][

ml-agents-envs/mlagents_envs/tests/test_set_action.py

@@ -0,0 +1,55 @@
+from mlagents_envs.registry import default_registry
+from mlagents_envs.side_channel.engine_configuration_channel import (
+    EngineConfigurationChannel,
+)
+from mlagents_envs.base_env import ActionTuple
+import numpy as np
+
+BALL_ID = "3DBall"
+
+
+def test_set_action_single_agent():
+    engine_config_channel = EngineConfigurationChannel()
+    env = default_registry[BALL_ID].make(
+        base_port=6000,
+        worker_id=0,
+        no_graphics=True,
+        side_channels=[engine_config_channel],
+    )
+    engine_config_channel.set_configuration_parameters(time_scale=100)
+    for _ in range(3):
+        env.reset()
+        behavior_name = list(env.behavior_specs.keys())[0]
+        d, t = env.get_steps(behavior_name)
+        for _ in range(50):
+            for agent_id in d.agent_id:
+                action = np.ones((1, 2))
+                action_tuple = ActionTuple()
+                action_tuple.add_continuous(action)
+                env.set_action_for_agent(behavior_name, agent_id, action_tuple)
+            env.step()
+            d, t = env.get_steps(behavior_name)
+    env.close()
+
+
+def test_set_action_multi_agent():
+    engine_config_channel = EngineConfigurationChannel()
+    env = default_registry[BALL_ID].make(
+        base_port=6001,
+        worker_id=0,
+        no_graphics=True,
+        side_channels=[engine_config_channel],
+    )
+    engine_config_channel.set_configuration_parameters(time_scale=100)
+    for _ in range(3):
+        env.reset()
+        behavior_name = list(env.behavior_specs.keys())[0]
+        d, t = env.get_steps(behavior_name)
+        for _ in range(50):
+            action = np.ones((len(d), 2))
+            action_tuple = ActionTuple()
+            action_tuple.add_continuous(action)
+            env.set_actions(behavior_name, action_tuple)
+            env.step()
+            d, t = env.get_steps(behavior_name)
+    env.close()