diff --git a/README.md b/README.md
index 22e9646..cd7b010 100644
--- a/README.md
+++ b/README.md
@@ -34,13 +34,14 @@ The following is an indicative list of examples.
- REINFORCE algorithm on ```CartPole-v0```
- Expected SARSA on ```CliffWalking-v0```
- Approximate Monte Carlo on ```MountainCar-v0```
-- Monte Carlo tree search on ```Taxi-v3```
+- Monte Carlo tree search on ```Taxi-v3```
- Double Q-learning on ```CartPole-v0```
- First visit Monte Carlo on ```FrozenLake-v0```
### Filtering
- Kalman filtering
+- Extended Kalman filter for diff-drive system
### Path planning
diff --git a/examples/filtering/CMakeLists.txt b/examples/filtering/CMakeLists.txt
index 07a51d4..130e5d0 100644
--- a/examples/filtering/CMakeLists.txt
+++ b/examples/filtering/CMakeLists.txt
@@ -1,2 +1,3 @@
ADD_SUBDIRECTORY(filtering_example_1)
+ADD_SUBDIRECTORY(filtering_example_2)
diff --git a/examples/filtering/filtering_example_2/CMakeLists.txt b/examples/filtering/filtering_example_2/CMakeLists.txt
new file mode 100644
index 0000000..c09580f
--- /dev/null
+++ b/examples/filtering/filtering_example_2/CMakeLists.txt
@@ -0,0 +1,18 @@
+cmake_minimum_required(VERSION 3.6 FATAL_ERROR)
+
+SET(EXECUTABLE filtering_example_2)
+SET(SOURCE ${EXECUTABLE}.cpp)
+
+ADD_EXECUTABLE(${EXECUTABLE} ${SOURCE})
+
+TARGET_LINK_LIBRARIES(${EXECUTABLE} cubeailib)
+
+IF( USE_PYTORCH )
+ TARGET_LINK_LIBRARIES(${EXECUTABLE} ${TORCH_LIBRARIES})
+ENDIF()
+
+TARGET_LINK_LIBRARIES(${EXECUTABLE} pthread)
+TARGET_LINK_LIBRARIES(${EXECUTABLE} openblas)
+TARGET_LINK_LIBRARIES(${EXECUTABLE} rlenvscpplib)
+target_link_libraries(${EXECUTABLE} tbb)
+target_link_libraries(${EXECUTABLE} boost_log)
diff --git a/examples/filtering/filtering_example_2/filtering_example_2.cpp b/examples/filtering/filtering_example_2/filtering_example_2.cpp
new file mode 100755
index 0000000..eee917d
--- /dev/null
+++ b/examples/filtering/filtering_example_2/filtering_example_2.cpp
@@ -0,0 +1,222 @@
+#include "cubeai/base/cubeai_types.h"
+#include "cubeai/estimation/extended_kalman_filter.h"
+#include "cubeai/utils/iteration_counter.h"
+#include "cubeai/io/csv_file_writer.h"
+#include "rlenvs/dynamics/diff_drive_dynamics.h"
+#include "rlenvs/dynamics/system_state.h"
+#include "rlenvs/utils/unit_converter.h"
+
+#include
+#include
+#include
+#include
+#include
+#include
+
+namespace example_2
+{
+
+using cubeai::real_t;
+using cubeai::uint_t;
+using cubeai::DynMat;
+using cubeai::DynVec;
+using cubeai::estimation::ExtendedKalmanFilter;
+using cubeai::utils::IterationCounter;
+using cubeai::io::CSVWriter;
+using rlenvscpp::dynamics::DiffDriveDynamics;
+using rlenvscpp::dynamics::SysState;
+
+
+class ObservationModel
+{
+
+public:
+
+ // the ExtendedKalmanFilter expects an exposed
+ // input_type
+ typedef DynVec input_type;
+
+ ObservationModel();
+
+ // simply return the state
+ const DynVec evaluate(const DynVec& input)const;
+
+ // get the H or M matrix
+ const DynMat& get_matrix(const std::string& name)const;
+
+private:
+
+ DynMat H;
+ DynMat M;
+};
+
+ObservationModel::ObservationModel()
+ :
+ H(2, 3),
+ M(2, 2)
+{
+ H = DynMat::Zero(2,3);
+ M = DynMat::Zero(2,2);
+ H(0, 0) = 1.0;
+ H(1,1) = 1.0;
+ M(0,0) = 1.0;
+ M(1, 1) = 1.0;
+
+}
+
+const DynVec
+ObservationModel::evaluate(const DynVec& input)const{
+ return input;
+}
+
+const DynMat&
+ObservationModel::get_matrix(const std::string& name)const{
+ if(name == "H"){
+ return H;
+ }
+ else if(name == "M"){
+ return M;
+ }
+
+ throw std::logic_error("Invalid matrix name. Name "+name+ " not found");
+}
+
+const DynVec get_measurement(const SysState<3>& state){
+
+ DynVec result(2);
+ result[0] = state.get("X");
+ result[1] = state.get("Y");
+ return result;
+}
+
+
+}
+
+int main() {
+
+ using namespace example_2;
+
+ BOOST_LOG_TRIVIAL(info)<<"Starting example...";
+
+ uint_t n_steps = 300;
+ auto time = 0.0;
+ auto dt = 0.5;
+
+ /// angular velocity
+ auto w = 0.0;
+
+ /// linear velocity
+ auto vt = 1.0;
+
+ std::array motion_control_error;
+ motion_control_error[0] = 0.0;
+ motion_control_error[1] = 0.0;
+
+ DiffDriveDynamics exact_motion_model;
+ exact_motion_model.set_matrix_update_flag(false);
+ exact_motion_model.set_time_step(dt);
+
+ DiffDriveDynamics motion_model;
+ motion_model.set_time_step(dt);
+
+ std::map input;
+ input["v"] = 1.0;
+ input["w"] = 0.0;
+ input["errors"] = motion_control_error;
+ motion_model.initialize_matrices(input);
+
+ ObservationModel observation;
+
+ ExtendedKalmanFilter ekf(motion_model, observation);
+
+ DynMat R = DynMat::Zero(2, 2);
+ R(0,0) = 1.0;
+ R(1, 1) = 1.0;
+
+ DynMat Q = DynMat::Zero(2, 2);
+ Q(0,0) = 0.001;
+ Q(1, 1) = 0.001;
+
+ DynMat P = DynMat::Zero(3, 3);
+ P(0, 0) = 1.0;
+ P(1, 1) = 1.0;
+ P(2, 2) = 1.0;
+
+ ekf.with_matrix("P", P)
+ .with_matrix("R", R)
+ .with_matrix("Q", Q);
+
+ CSVWriter writer("state");
+ writer.open();
+ std::vector names{"Time", "X_true", "Y_true", "Theta_true", "X", "Y", "Theta"};
+ writer.write_column_names(names);
+
+ try{
+
+
+ BOOST_LOG_TRIVIAL(info)<<"Starting simulation... "< motion_input;
+ motion_input["v"] = vt; // we keep the velocity constant
+ motion_input["errors"] = motion_control_error;
+
+ for(uint_t step=0; step < n_steps; ++step){
+
+ BOOST_LOG_TRIVIAL(info)<<"Simulation time: "< row(7, 0.0);
+ row[0] = time;
+ row[1] = exact_state.get("X");
+ row[2] = exact_state.get("Y");
+ row[3] = exact_state.get("Theta");
+ row[4] = state.get("X");
+ row[5] = state.get("Y");
+ row[6] = state.get("Theta");
+ writer.write_row(row);
+
+ time += dt;
+ counter++;
+ }
+
+ BOOST_LOG_TRIVIAL(info)<<"Finished example...";
+ }
+ catch(std::runtime_error& e){
+ std::cerr<
+#include
+#include
+#include
+#include
+#include
+
+namespace example_2
+{
+
+using cubeai::real_t;
+using cubeai::uint_t;
+using cubeai::DynMat;
+using cubeai::DynVec;
+using cubeai::estimation::ExtendedKalmanFilter;
+using cubeai::utils::IterationCounter;
+using cubeai::io::CSVWriter;
+using rlenvscpp::dynamics::DiffDriveDynamics;
+using rlenvscpp::dynamics::SysState;
+
+
+class ObservationModel
+{
+
+public:
+
+ // the ExtendedKalmanFilter expects an exposed
+ // input_type
+ typedef DynVec input_type;
+
+ ObservationModel();
+
+ // simply return the state
+ const DynVec evaluate(const DynVec& input)const;
+
+ // get the H or M matrix
+ const DynMat& get_matrix(const std::string& name)const;
+
+private:
+
+ DynMat H;
+ DynMat M;
+};
+
+ObservationModel::ObservationModel()
+ :
+ H(2, 3),
+ M(2, 2)
+{
+ H = DynMat::Zero(2,3);
+ M = DynMat::Zero(2,2);
+ H(0, 0) = 1.0;
+ H(1,1) = 1.0;
+ M(0,0) = 1.0;
+ M(1, 1) = 1.0;
+
+}
+
+const DynVec
+ObservationModel::evaluate(const DynVec& input)const{
+ return input;
+}
+
+const DynMat&
+ObservationModel::get_matrix(const std::string& name)const{
+ if(name == "H"){
+ return H;
+ }
+ else if(name == "M"){
+ return M;
+ }
+
+ throw std::logic_error("Invalid matrix name. Name "+name+ " not found");
+}
+
+const DynVec get_measurement(const SysState<3>& state){
+
+ DynVec result(2);
+ result[0] = state.get("X");
+ result[1] = state.get("Y");
+ return result;
+}
+
+
+}
+
+int main() {
+
+ using namespace example_2;
+
+ BOOST_LOG_TRIVIAL(info)<<"Starting example...";
+
+ uint_t n_steps = 300;
+ auto time = 0.0;
+ auto dt = 0.5;
+
+ /// angular velocity
+ auto w = 0.0;
+
+ /// linear velocity
+ auto vt = 1.0;
+
+ std::array motion_control_error;
+ motion_control_error[0] = 0.0;
+ motion_control_error[1] = 0.0;
+
+ DiffDriveDynamics exact_motion_model;
+ exact_motion_model.set_matrix_update_flag(false);
+ exact_motion_model.set_time_step(dt);
+
+ DiffDriveDynamics motion_model;
+ motion_model.set_time_step(dt);
+
+ std::map input;
+ input["v"] = 1.0;
+ input["w"] = 0.0;
+ input["errors"] = motion_control_error;
+ motion_model.initialize_matrices(input);
+
+ ObservationModel observation;
+
+ ExtendedKalmanFilter ekf(motion_model, observation);
+
+ DynMat R = DynMat::Zero(2, 2);
+ R(0,0) = 1.0;
+ R(1, 1) = 1.0;
+
+ DynMat Q = DynMat::Zero(2, 2);
+ Q(0,0) = 0.001;
+ Q(1, 1) = 0.001;
+
+ DynMat P = DynMat::Zero(3, 3);
+ P(0, 0) = 1.0;
+ P(1, 1) = 1.0;
+ P(2, 2) = 1.0;
+
+ ekf.with_matrix("P", P)
+ .with_matrix("R", R)
+ .with_matrix("Q", Q);
+
+ CSVWriter writer("state");
+ writer.open();
+ std::vector names{"Time", "X_true", "Y_true", "Theta_true", "X", "Y", "Theta"};
+ writer.write_column_names(names);
+
+ try{
+
+
+ BOOST_LOG_TRIVIAL(info)<<"Starting simulation... "< motion_input;
+ motion_input["v"] = vt; // we keep the velocity constant
+ motion_input["errors"] = motion_control_error;
+
+ for(uint_t step=0; step < n_steps; ++step){
+
+ BOOST_LOG_TRIVIAL(info)<<"Simulation time: "< row(7, 0.0);
+ row[0] = time;
+ row[1] = exact_state.get("X");
+ row[2] = exact_state.get("Y");
+ row[3] = exact_state.get("Theta");
+ row[4] = state.get("X");
+ row[5] = state.get("Y");
+ row[6] = state.get("Theta");
+ writer.write_row(row);
+
+ time += dt;
+ counter++;
+ }
+
+ BOOST_LOG_TRIVIAL(info)<<"Finished example...";
+ }
+ catch(std::runtime_error& e){
+ std::cerr<
diff --git a/examples/rl/rl_example_10/rl_example_10.cpp b/examples/rl/rl_example_10/rl_example_10.cpp
index de91511..a19f5ba 100755
--- a/examples/rl/rl_example_10/rl_example_10.cpp
+++ b/examples/rl/rl_example_10/rl_example_10.cpp
@@ -25,8 +25,8 @@ using cubeai::rl::algos::td::QLearningConfig;
using cubeai::rl::policies::EpsilonDecayOption;
using cubeai::rl::RLSerialAgentTrainer;
using cubeai::rl::RLSerialTrainerConfig;
-using rlenvs_cpp::envs::gymnasium::CliffWorldActionsEnum;
-typedef rlenvs_cpp::envs::gymnasium::CliffWorld env_type;
+
+typedef rlenvscpp::envs::gymnasium::CliffWorld env_type;
}
diff --git a/examples/rl/rl_example_11/rl_example_11.cpp b/examples/rl/rl_example_11/rl_example_11.cpp
index 329f86a..37d39d6 100755
--- a/examples/rl/rl_example_11/rl_example_11.cpp
+++ b/examples/rl/rl_example_11/rl_example_11.cpp
@@ -28,7 +28,6 @@ using cubeai::rl::algos::ac::A2CConfig;
using cubeai::rl::algos::ac::A2CSolver;
using cubeai::rl::RLSerialAgentTrainer;
using cubeai::rl::RLSerialTrainerConfig;
-using rlenvs_cpp::envs::gymnasium::CartPoleActionsEnum;
// create the Action and the Critic networks
@@ -132,7 +131,7 @@ CriticNetImpl::forward(torch_tensor_t state){
TORCH_MODULE(ActorNet);
TORCH_MODULE(CriticNet);
-typedef rlenvs_cpp::envs::gymnasium::CartPole env_type;
+typedef rlenvscpp::envs::gymnasium::CartPole env_type;
}
diff --git a/examples/rl/rl_example_12/rl_example_12.cpp b/examples/rl/rl_example_12/rl_example_12.cpp
index 63e1986..ad2e530 100755
--- a/examples/rl/rl_example_12/rl_example_12.cpp
+++ b/examples/rl/rl_example_12/rl_example_12.cpp
@@ -41,7 +41,7 @@ using cubeai::torch_tensor_t;
using cubeai::rl::RLSerialAgentTrainer;
using cubeai::rl::RLSerialTrainerConfig;
using cubeai::rl::policies::EpsilonGreedyPolicy;
-using rlenvs_cpp::envs::grid_world::Gridworld;
+using rlenvscpp::envs::grid_world::Gridworld;
const uint_t l1 = 64;
const uint_t l2 = 150;
diff --git a/examples/rl/rl_example_13/rl_example_13.cpp b/examples/rl/rl_example_13/rl_example_13.cpp
index 76f4ca7..887dd59 100755
--- a/examples/rl/rl_example_13/rl_example_13.cpp
+++ b/examples/rl/rl_example_13/rl_example_13.cpp
@@ -43,7 +43,7 @@ using cubeai::rl::algos::pg::ReinforceConfig;
using cubeai::rl::RLSerialAgentTrainer;
using cubeai::rl::RLSerialTrainerConfig;
using cubeai::maths::stats::TorchCategorical;
-using rlenvs_cpp::envs::gymnasium::CartPole;
+using rlenvscpp::envs::gymnasium::CartPole;
const uint_t L1 = 4;
diff --git a/examples/rl/rl_example_15/rl_example_15.cpp b/examples/rl/rl_example_15/rl_example_15.cpp
index ea7360e..2940880 100755
--- a/examples/rl/rl_example_15/rl_example_15.cpp
+++ b/examples/rl/rl_example_15/rl_example_15.cpp
@@ -42,7 +42,7 @@ using cubeai::rl::RLSerialAgentTrainer;
using cubeai::rl::RLSerialTrainerConfig;
using cubeai::rl::policies::EpsilonGreedyPolicy;
using cubeai::containers::ExperienceBuffer;
-using rlenvs_cpp::envs::grid_world::Gridworld;
+using rlenvscpp::envs::grid_world::Gridworld;
const std::string EXPERIMENT_ID = "1";
@@ -176,7 +176,7 @@ int main(){
// initialize the environment using random mode
std::unordered_map options;
- options["mode"] = std::any(rlenvs_cpp::envs::grid_world::to_string(rlenvs_cpp::envs::grid_world::GridWorldInitType::RANDOM));
+ options["mode"] = std::any(rlenvscpp::envs::grid_world::to_string(rlenvscpp::envs::grid_world::GridWorldInitType::RANDOM));
env.make("v0", options);
diff --git a/examples/rl/rl_example_16/CMakeLists.txt b/examples/rl/rl_example_16/CMakeLists.txt
new file mode 100644
index 0000000..ee2e4ca
--- /dev/null
+++ b/examples/rl/rl_example_16/CMakeLists.txt
@@ -0,0 +1,19 @@
+cmake_minimum_required(VERSION 3.6 FATAL_ERROR)
+
+SET(EXECUTABLE rl_example_16)
+SET(SOURCE ${EXECUTABLE}.cpp)
+
+ADD_EXECUTABLE(${EXECUTABLE} ${SOURCE})
+
+TARGET_LINK_LIBRARIES(${EXECUTABLE} cubeailib)
+
+IF( USE_PYTORCH )
+ TARGET_LINK_LIBRARIES(${EXECUTABLE} ${TORCH_LIBRARIES})
+ENDIF()
+
+TARGET_LINK_LIBRARIES(${EXECUTABLE} pthread)
+TARGET_LINK_LIBRARIES(${EXECUTABLE} openblas)
+TARGET_LINK_LIBRARIES(${EXECUTABLE} rlenvscpplib)
+target_link_libraries(${EXECUTABLE} tbb)
+target_link_libraries(${EXECUTABLE} boost_log)
+
diff --git a/examples/rl/example_16/example_16.cpp b/examples/rl/rl_example_16/rl_example_16.cpp
similarity index 86%
rename from examples/rl/example_16/example_16.cpp
rename to examples/rl/rl_example_16/rl_example_16.cpp
index e1abc35..95ad42d 100755
--- a/examples/rl/example_16/example_16.cpp
+++ b/examples/rl/rl_example_16/rl_example_16.cpp
@@ -8,15 +8,9 @@
#include "cubeai/base/cubeai_types.h"
-#include "cubeai/rl/algorithms/mc/mc_tree_search_base.h"
-#include "cubeai/utils/array_utils.h"
+//#include "cubeai/rl/algorithms/mc/mc_tree_search_solver.h"
-#include "gymfcpp/gymfcpp_types.h"
-#include "gymfcpp/taxi_env.h"
-#include "gymfcpp/time_step.h"
-#include "gymfcpp/gymfcpp_consts.h"
-
-#include
+#include "rlenvs/envs/connect2/connect2_env.h"
#include
#include
@@ -28,12 +22,13 @@
namespace example{
+ /*
using cubeai::real_t;
using cubeai::uint_t;
using cubeai::rl::algos::MCTreeSearchBase;
using cubeai::rl::algos::MCTreeSearchConfig;
-using gymfcpp::Taxi;
+using rlenvscpp::envs::connect2::Connect2;
typedef Taxi::time_step_type time_step_type;
@@ -43,10 +38,7 @@ const uint_t N_EPISODES = 20000;
const uint_t N_ITRS_PER_EPISODE = 2000;
const real_t TOL = 1.0e-8;
-/**
- * Implementation of Monte Carlo tree serach for the OpenAI-Gym
- * environment.
- */
+
template
class TaxiMCTreeSearch: public MCTreeSearchBase
{
@@ -115,7 +107,8 @@ TaxiMCTreeSearch::simulate_node(std::shared_ptr node, env_type&
return time_step;
}
-
+*/
+/*
template
void
TaxiMCTreeSearch::on_episode(){
@@ -188,7 +181,9 @@ TaxiMCTreeSearch::on_episode(){
best_rewards_.push_back(sum_reward_);
}
+*/
+/*
template
void
TaxiMCTreeSearch::expand_node(std::shared_ptr node, env_type& env){
@@ -214,6 +209,7 @@ TaxiMCTreeSearch::backprop(std::shared_ptr node){
}
+ */
}
@@ -223,23 +219,8 @@ int main(){
try{
- Py_Initialize();
- auto main_module = boost::python::import("__main__");
- auto main_namespace = main_module.attr("__dict__");
-
- Taxi env("v0", main_namespace, false);
- env.make();
+
- MCTreeSearchConfig config;
- config.max_tree_depth = 512;
- config.temperature = 1.0;
- TaxiMCTreeSearch agent(config, env);
- agent.train();
-
- }
- catch(const boost::python::error_already_set&)
- {
- PyErr_Print();
}
catch(std::exception& e){
std::cout<Monte Carlo tree search algorithm
+or MCTS for short.
+
+----
+**Remark**
+
+A concise presentation of the algorithm can be found in the following video: https://www.youtube.com/watch?v=UXW2yZndl7U
+
+The following videos from Udacity's Reinforcement Learning also explain
+the MCTS algorithm
+
+1. Monte Carlo Tree Search p1
+2. Monte Carlo Tree Search p2
+3. Monte Carlo Tree Search p3
+4. Monte Carlo Tree Search p4
+
+
+In this tutorial we will be applying MCTS in a toy environment namely Connect2. The following video is a nice discussion
+on the application of MCTS on Connect2: Alpha Zero and Monte Carlo Tree Search.
+
+----
+
+Overall, the vanilla MCTS algorithm has four steps
+
+1. Tree tranversal
+2. Node expansion
+3. Rollout or random simulation
+4. Backward propagation
+
+Let's go over each one of these steps.
+
+### Tree traversal
+
+At this step the agent select the root node from which it will recursively select the best paths.
+This is done until a leaf node is reached. The selction of a node is typically done using the
+UCB (or upper confidence bound) criterion
+but you can use any criterion that is suitable for your application but make sure that the criterion you are using
+balances exploration and exploitation. The UCB criterion is as follows
+
+
+
+Where $N$ is the number of node visits
+The UCB criterion limits the extent that each each branch will be explored.
+This is done because as $N$ increases beacuse we visit the node more and more,
+the overall factor continuously decreases thus the expected reward from the current branch will also decrease.
+
+A new node is created when we execute the rollout step.
+
+
+
+### Rollout step
+
+The MCTS starts with an empty tree data structure and iteratively will build a portion of the tree by running a number of simulations
+Each simulation will add a single node to the tree. Since a simulation is the exploration step, the more simualtions we run, the beter
+we expect the model to perform. However, simulating indefinitetly is not possible for obvious reasons.
+
+The MCTS algorithm implies that somehow we have a way of doing the simulation step. One way is if we know the transition dynamics of the
+environment or being able to sample from the environment.
+
+### Backward propagation
+
+During the backward step we update the statistics for every node in the search path. These statistics affect the UCB score and will
+guide the next MCTS simulation.
+
+Now that we have gone over the detaisl of the MCTS algorithm let's try to answer some practical questions. Fo example when does the algorithm stop?
+There are various ways to address this question and the most common one is to simulate up until a specific number of iterations.
+
+All in all, the MCTS algorithm is summarised below.
+
+In this
+
+## Driver code
+
+
+## References
+
+
+
+
+
diff --git a/examples/rl/rl_example_19/rl_example_19.cpp b/examples/rl/rl_example_19/rl_example_19.cpp
index 08ffdcd..7b2724d 100755
--- a/examples/rl/rl_example_19/rl_example_19.cpp
+++ b/examples/rl/rl_example_19/rl_example_19.cpp
@@ -29,8 +29,8 @@ using cubeai::rl::algos::mc::FirstVisitMCSolver;
using cubeai::rl::algos::mc::FirstVisitMCSolverConfig;
using cubeai::rl::RLSerialAgentTrainer;
using cubeai::rl::RLSerialTrainerConfig;
-using rlenvs_cpp::envs::gymnasium::FrozenLake;
-using rlenvs_cpp::envs::gymnasium::FrozenLakeActionsEnum;
+using rlenvscpp::envs::gymnasium::FrozenLake;
+using rlenvscpp::envs::gymnasium::FrozenLakeActionsEnum;
const std::string SERVER_URL = "http://0.0.0.0:8001/api";
const uint_t SEED = 42;
@@ -78,7 +78,7 @@ std::vector::time_step_type>
TrajectoryGenerator::operator()(FrozenLake<4>& env, uint_t max_steps){
RandomActionSelector action_selector;
- return rlenvs_cpp::envs::create_trajectory(env, action_selector, max_steps );
+ return rlenvscpp::envs::create_trajectory(env, action_selector, max_steps );
}
diff --git a/examples/rl/rl_example_6/rl_example_6.cpp b/examples/rl/rl_example_6/rl_example_6.cpp
index a8a0263..8ba542e 100755
--- a/examples/rl/rl_example_6/rl_example_6.cpp
+++ b/examples/rl/rl_example_6/rl_example_6.cpp
@@ -25,7 +25,7 @@ using cubeai::rl::algos::dp::IterativePolicyEvalutationSolver;
using cubeai::rl::algos::dp::IterativePolicyEvalConfig;
using cubeai::rl::RLSerialAgentTrainer;
using cubeai::rl::RLSerialTrainerConfig;
-using rlenvs_cpp::envs::gymnasium::FrozenLake;
+using rlenvscpp::envs::gymnasium::FrozenLake;
typedef FrozenLake<4> env_type;
}
diff --git a/examples/rl/rl_example_7/rl_example_7.cpp b/examples/rl/rl_example_7/rl_example_7.cpp
index a084c01..1c45b77 100755
--- a/examples/rl/rl_example_7/rl_example_7.cpp
+++ b/examples/rl/rl_example_7/rl_example_7.cpp
@@ -30,7 +30,7 @@ using cubeai::rl::algos::dp::PolicyIterationSolver;
using cubeai::rl::algos::dp::PolicyIterationConfig;
using cubeai::rl::RLSerialAgentTrainer;
using cubeai::rl::RLSerialTrainerConfig;
-using rlenvs_cpp::envs::gymnasium::FrozenLake;
+using rlenvscpp::envs::gymnasium::FrozenLake;
typedef FrozenLake<4> env_type;
}
diff --git a/examples/rl/rl_example_8/rl_example_8.cpp b/examples/rl/rl_example_8/rl_example_8.cpp
index d58b983..3657825 100755
--- a/examples/rl/rl_example_8/rl_example_8.cpp
+++ b/examples/rl/rl_example_8/rl_example_8.cpp
@@ -23,7 +23,7 @@ using cubeai::rl::algos::dp::ValueIterationConfig;
using cubeai::rl::RLSerialAgentTrainer;
using cubeai::rl::RLSerialTrainerConfig;
-using rlenvs_cpp::envs::gymnasium::FrozenLake;
+using rlenvscpp::envs::gymnasium::FrozenLake;
typedef FrozenLake<4> env_type;
typedef ValueIteration;
///
+ /// Float type vector
///
- ///
- using FloatVec = DynVec;
+ using FloatVec = DynVec;
+
+ ///
+ /// Real type vector
+ ///
+ using RealVec = DynVec;
///
/// \brief Null type. Simple placeholder
diff --git a/include/cubeai/estimation/extended_kalman_filter.h b/include/cubeai/estimation/extended_kalman_filter.h
new file mode 100644
index 0000000..032eb85
--- /dev/null
+++ b/include/cubeai/estimation/extended_kalman_filter.h
@@ -0,0 +1,324 @@
+#ifndef EXTENDED_KALMAN_FILTER_H
+#define EXTENDED_KALMAN_FILTER_H
+
+#include "cubeai/base/cubeai_types.h"
+
+#include
+#include