started with bo node

src/interaction_msgs/CMakeLists.txt

@@ -10,6 +10,10 @@ find_package(rosidl_default_generators REQUIRED)
 
 rosidl_generate_interfaces(${PROJECT_NAME}
     "srv/Query.srv"
+    "msg/Opt2Rep.msg"
+    "msg/Opt2UI.msg"
+    "msg/Task2Opt.msg"
+    "msg/UI2Opt.msg"
     DEPENDENCIES
 )
 

src/interaction_msgs/msg/Opt2Rep.msg

@@ -0,0 +1,6 @@
+# Number of dimensions of the representation model
+uint16 nr_dim
+# Number of parameters per dimensions
+uint16 nr_parameter
+# Next parameters
+float32[] x_next

src/interaction_msgs/msg/Opt2UI.msg

@@ -0,0 +1,5 @@
+# Best parameter set so far
+float32[] x_best
+
+# parameters which were fixed by the user
+bool[] fixed_parameters

src/interaction_msgs/msg/Task2Opt.msg

@@ -0,0 +1,4 @@
+# observed parameters
+float32[] x_observed
+# observed reward
+float32 reward

src/interaction_msgs/msg/UI2Opt.msg

@@ -0,0 +1,4 @@
+# parameters which were fixed by the user
+bool[] fixed_parameters
+# parameters set by the user
+float32[] set_parameter_values

src/interaction_optimizers/interaction_optimizers/bayesian_optimization_node.py

@@ -1 +1,56 @@
 
+import rclpy
+from rclpy.node import Node
+
+from rclpy.callback_groups import ReentrantCallbackGroup
+
+from interaction_msgs.srv import Query
+
+from .optimizers.bayesian_optimization import BayesianOptimization
+
+import numpy as np
+import time
+import os
+
+
+class BayesianOptimizationNode(Node):
+    def __init__(self):
+        super().__init__('bayesian_optimization_node')
+
+        # reading parameters
+        self.kernel_type = self.declare_parameter('kernel_type', 'Matern').get_parameter_value().string_value
+        self.acquisition_function_name = (self.declare_parameter('acquisition_function_name', 'EI')
+                                     .get_parameter_value().string_value)
+
+        # Subscribers
+
+        # Publishers
+
+        # Service Clients
+        self.query_client = self.create_client(Query, 'user_query')
+
+        # Bayesian Optimization
+        self.bayesian_optimization = None
+        self.nr_bo_steps = 100
+        self.nr_dimensions = 1
+        self.nr_policy_parameters = 10
+
+        self.nr_init = 3
+        self.seed = None
+        self.lower_bound = None
+        self.upper_bound = None
+
+    def reset_bo(self, fixed_dimensions=None, **kwargs):
+        self.bayesian_optimization = BayesianOptimization(self.nr_bo_steps,
+                                                          self.nr_dimensions,
+                                                          self.nr_policy_parameters,
+                                                          seed=self.seed,
+                                                          fixed_dimensions=fixed_dimensions,
+                                                          lower_bound=self.lower_bound,
+                                                          upper_bound=self.upper_bound,
+                                                          acquisition_function_name=self.acquisition_function_name,
+                                                          kernel_type=self.kernel_type, **kwargs)
+
+
+
+