[python/plot] More versatile plotting tools. (#302)

* [core|python] Add support of non-contiguous temporary to 'register_variables'. 
* [core/python] Allow registering 2D numpy arrays for bindings.
* [python/processing] Add N-order integrator assuming zero-order-hold for highest derivative.
* [python|gym] Add TabbedFigure class and use it in simulator. 
* [python|gym] Move '_log_data' from gym env to simulator. Log generic action automatically.
* [misc] Remove force install 'numba==0.53.0rc3' on CI since 0.53 is now available.
* [misc] Consistent use of 'resource_filename' to access models data.
* [misc] Add Atlas in Readme instead of Anymal.
* [misc] Fix ipython tutorial.
* [python/plot] Add option to sync tabs. Fix remove tab and clear figure.
* [python/simulator] Fix log backup cleanup.
* [python/simulator] Enable argument forwarding for 'plot' method.
* [misc] Fix type checker errors.
* [python/dynamics] Fix 'update_quantities' not calling forward kinematics when necessary. Compute everything between jacobians by default.
* [python/plot] Only refresh canvas when needed. 
* [python/plot] Automatically re-arrange subplot grid. 
* [python/plot] Add export methods. 
* [python/plot] Fix support of any backend. 
* [python/plot] Add partial documentation.
* [python/plot] Fix buttons responsiveness and active tab hovering color.
* [python/plot] Disable click button for active tab.
* [python/plot] Add 'close' method. 
* [python/plot] Add 'image_path' option argument to 'plot' method.
* [gym/common/pipeline] Fix controller wrapper not updating env action and not calling env 'compute_command'.

Co-authored-by: Alexis Duburcq <[email protected]>

.github/workflows/linux.yml

@@ -53,7 +53,7 @@ jobs:
       run: |
         ./build_tools/build_install_deps_linux.sh
         "${PYTHON_EXECUTABLE}" -m pip install torch==1.8.0+cpu -f https://download.pytorch.org/whl/torch_stable.html
-        "${PYTHON_EXECUTABLE}" -m pip install "gym>=0.17.3,<0.18.0" "numba==0.53.0rc3" "stable_baselines3>=0.10"
+        "${PYTHON_EXECUTABLE}" -m pip install "gym>=0.17.3,<0.18.0" "stable_baselines3>=0.10"
 
         PYTHON_USER_SITELIB="$("${PYTHON_EXECUTABLE}" -m site --user-site)"
         mkdir -p "${PYTHON_USER_SITELIB}"

.github/workflows/manylinux.yml

@@ -57,7 +57,7 @@ jobs:
     - name: Build project dependencies
       run: |
         ./build_tools/build_install_deps_linux.sh
-        "${PYTHON_EXECUTABLE}" -m pip install "gym>=0.17.3,<0.18.0" "numba==0.53.0rc3" "stable_baselines3>=0.10"
+        "${PYTHON_EXECUTABLE}" -m pip install "gym>=0.17.3,<0.18.0" "stable_baselines3>=0.10"
 
     #####################################################################################
 

.github/workflows/ubuntu.yml

@@ -48,7 +48,7 @@ jobs:
       run: |
         sudo "${GITHUB_WORKSPACE}/build_tools/easy_install_deps_ubuntu.sh"
         "${PYTHON_EXECUTABLE}" -m pip install "torch==1.8.0+cpu" -f https://download.pytorch.org/whl/torch_stable.html
-        "${PYTHON_EXECUTABLE}" -m pip install "gym>=0.17.3,<0.18.0" "numba==0.53.0rc3" "stable_baselines3>=0.10"
+        "${PYTHON_EXECUTABLE}" -m pip install "gym>=0.17.3,<0.18.0" "stable_baselines3>=0.10"
 
     #####################################################################################
 

.github/workflows/win.yml

@@ -47,7 +47,7 @@ jobs:
     - name: Build project dependencies
       run: |
         python -m pip install "torch==1.8.0+cpu" -f https://download.pytorch.org/whl/torch_stable.html
-        python -m pip install "gym>=0.17.3,<0.18.0" "numba==0.53.0rc3" "stable_baselines3>=0.10"
+        python -m pip install "gym>=0.17.3,<0.18.0" "stable_baselines3>=0.10"
         & "./build_tools/build_install_deps_windows.ps1"
 
     #####################################################################################

CMakeLists.txt

@@ -2,7 +2,7 @@
 cmake_minimum_required(VERSION 3.10)
 
 # Set the build version
-set(BUILD_VERSION 1.6.4)
+set(BUILD_VERSION 1.6.5)
 
 # Add definition of Jiminy version for C++ headers
 add_definitions("-DJIMINY_VERSION=\"${BUILD_VERSION}\"")

README.md

@@ -59,7 +59,7 @@ Pre-configured environments for some well-known toys models and reference roboti
   <img src="https://raw.github.com/duburcqa/jiminy/readme/jiminy_tensorboard_cartpole.png" alt="" width="98.5%"/>
   <img src="https://raw.github.com/duburcqa/jiminy/readme/jiminy_learning_ant.gif" alt="" width="32.5%"/>
   <img src="https://raw.github.com/duburcqa/jiminy/readme/cassie.png" alt="" width="32.5%"/>
-  <img src="https://raw.github.com/duburcqa/jiminy/readme/anymal.png" alt="" width="32.5%"/>
+  <img src="https://raw.github.com/duburcqa/jiminy/readme/atlas.png" alt="" width="32.5%"/>
 </p>
 </a>
 

core/include/jiminy/core/control/AbstractController.h

@@ -84,13 +84,13 @@ namespace jiminy
         ///             update it manually whenever it is necessary to do so.
         ///
         /// \param[in]  fieldnames      Name of each element of the variable. It will appear in the header of the log.
-        /// \param[in]  values          Eigen vector to add to the telemetry
+        /// \param[in]  values          Eigen vector to add to the telemetry. It accepts non-contiguous temporary.
         ///
         /// \return     Return code to determine whether the execution of the method was successful.
         ///
         ///////////////////////////////////////////////////////////////////////////////////////////////
         hresult_t registerVariable(std::vector<std::string> const & fieldnames,
-                                   Eigen::Ref<vectorN_t>            values);  // Make a "copy" to support both vectorN_t and reference
+                                   Eigen::Ref<vectorN_t, 0, Eigen::InnerStride<> > values);  // Make a "copy" to support both vectorN_t and reference
 
         ///////////////////////////////////////////////////////////////////////////////////////////////
         ///

core/include/jiminy/core/control/AbstractController.tpp

@@ -15,19 +15,20 @@ namespace jiminy
 {
     using std::to_string;
 
-    inline std::string to_string(char_t const * var)
-    {
-        return {var};
-    }
-
-    inline std::string to_string(Eigen::Ref<matrixN_t const> const & var)
+    template<typename DerivedType>
+    std::string to_string(Eigen::MatrixBase<DerivedType> const & var)
     {
         Eigen::IOFormat HeavyFmt(Eigen::FullPrecision, 0, ", ", ";\n", "[", "]", "[", "]");
         std::stringstream matrixStream;
         matrixStream << var.format(HeavyFmt);
         return matrixStream.str();
     }
 
+    inline std::string to_string(char_t const * var)
+    {
+        return {var};
+    }
+
     ///////////////////////////////////////////////////////////////////////////////////////////////
     ///
     /// \brief      Register a constant to the telemetry.

core/src/control/AbstractController.cc

@@ -167,7 +167,7 @@ namespace jiminy
     }
 
     hresult_t AbstractController::registerVariable(std::vector<std::string> const & fieldnames,
-                                                   Eigen::Ref<vectorN_t>            values)
+                                                   Eigen::Ref<vectorN_t, 0, Eigen::InnerStride<> > values)
     {
         // Delayed variable registration (Taken into account by 'configureTelemetry')
 
@@ -192,7 +192,7 @@ namespace jiminy
                 PRINT_ERROR("Variable already registered.");
                 return hresult_t::ERROR_BAD_INPUT;
             }
-            registeredVariables_.emplace_back(*fieldIt, values.data() + i);
+            registeredVariables_.emplace_back(*fieldIt, &values[i]);
         }
 
         return hresult_t::SUCCESS;

examples/tutorial.ipynb

@@ -60,36 +60,46 @@
    ],
    "source": [
     "import os\n",
+    "from pkg_resources import resource_filename\n",
+    "\n",
     "import numpy as np\n",
     "\n",
-    "import jiminy_py.core as jiminy # The main module of jiminy - this is what gives access to the Robot and Engine class.\n",
+    "import jiminy_py.core as jiminy  # The main module of jiminy - this is what gives access to the Robot\n",
+    "from jiminy_py.controller import BaseJiminyObserverController\n",
     "from jiminy_py.simulator import Simulator\n",
     "\n",
     "\n",
-    "data_root_path = \"../data/toys_models/simple_pendulum\"\n",
+    "data_root_path = resource_filename(\n",
+    "    \"gym_jiminy.envs\", \"data/toys_models/simple_pendulum\")\n",
     "urdf_path = os.path.join(data_root_path, \"simple_pendulum.urdf\")\n",
     "\n",
+    "# Instantiate and initialize the robot\n",
     "robot = jiminy.Robot()\n",
     "robot.initialize(urdf_path, mesh_package_dirs=[data_root_path])\n",
     "\n",
+    "# Add a single motor\n",
     "motor = jiminy.SimpleMotor(\"PendulumJoint\")\n",
     "robot.attach_motor(motor)\n",
     "motor.initialize(\"PendulumJoint\")\n",
     "\n",
-    "# Create the controller: for now, the motor is off and doesn't modify the output torque.\n",
+    "# Instantiate and initialize the controller\n",
+    "controller = BaseJiminyObserverController()\n",
+    "controller.initialize(robot)\n",
+    "\n",
+    "# Set the command: for now, the motor is off and doesn't modify the output torque.\n",
     "def compute_command(t, q, v, sensors_data, command):\n",
     "    command.fill(0.0)\n",
+    "controller.set_controller_handle(compute_command)\n",
     "\n",
     "# Create a simulator using this robot and controller\n",
-    "simulator = Simulator(robot)\n",
-    "simulator.controller.set_controller_handle(compute_command)\n",
+    "simulator = Simulator(robot, controller)\n",
     "\n",
     "# Set initial condition and simulation length\n",
-    "x0 = np.array([0.1, 0])\n",
+    "q0, v0 = np.array([0.1]), np.array([0.0])\n",
     "simulation_duration = 10.0\n",
     "\n",
     "# Launch the simulation\n",
-    "simulator.run(simulation_duration, x0)"
+    "simulator.run(simulation_duration, q0, v0)"
    ]
   },
   {
@@ -2693,8 +2703,8 @@
     "def compute_command(t, q, v, sensors_data, command):\n",
     "    command[:] = - Kp * (q + Kd * v)\n",
     "\n",
-    "simulator.controller.set_controller_handle(compute_command)\n",
-    "simulator.run(simulation_duration, x0)\n",
+    "controller.set_controller_handle(compute_command)\n",
+    "simulator.run(simulation_duration, q0, v0)\n",
     "\n",
     "log_data, _ = simulator.get_log()\n",
     "play_logfiles(robot, log_data, camera_xyzrpy=camera_xyzrpy)"
@@ -2740,7 +2750,7 @@
     "\n",
     "# Apply this force profile to a given frame.\n",
     "simulator.engine.register_force_profile(\"PendulumMass\", force_profile)\n",
-    "simulator.run(simulation_duration, x0)\n",
+    "simulator.run(simulation_duration, q0, v0)\n",
     "\n",
     "log_data, _ = simulator.get_log()\n",
     "play_logfiles(robot, log_data, camera_xyzrpy=camera_xyzrpy)"
@@ -2768,4 +2778,4 @@
  },
  "nbformat": 4,
  "nbformat_minor": 4
-}
+}

python/gym_jiminy/common/gym_jiminy/common/bases/block_bases.py

@@ -12,7 +12,7 @@
 
 import gym
 
-from ..utils import FieldDictNested, SpaceDictNested
+from ..utils import get_fieldnames, FieldDictNested, SpaceDictNested
 from ..envs import BaseJiminyEnv
 from .generic_bases import ControllerInterface, ObserverInterface
 
@@ -264,11 +264,14 @@ def get_fieldnames(self) -> FieldDictNested:
         action space, the difference being numerical arrays replaced by lists
         of string.
 
+        By default, generic fieldnames using 'Action' prefix and index as
+        suffix for `np.ndarray`.
+
         .. note::
             This method is not supposed to be called before `reset`, so that
             the controller should be already initialized at this point.
         """
-        raise NotImplementedError
+        return get_fieldnames(self.action_space)
 
 
 BaseControllerBlock._setup.__doc__ = \

python/gym_jiminy/common/gym_jiminy/common/bases/pipeline_bases.py

@@ -81,7 +81,7 @@ def _controller_handle(self,
                            sensors_data: jiminy.sensorsData,
                            command: np.ndarray) -> None:
         command[:] = self.compute_command(
-            self.env.get_observation(), self._action)
+            self.env.get_observation(), deepcopy(self._action))
 
     def _get_block_index(self) -> int:
         """Get the index of the block. It corresponds the "deepness" of the
@@ -219,6 +219,7 @@ def compute_command(self,
         :param action: Target to achieve.
         """
         set_value(self._action, action)
+        set_value(self.env._action, action)
         return self.env.compute_command(measure, action)
 
 
@@ -515,9 +516,10 @@ def _setup(self) -> None:
 
         # Register the controller target to the telemetry.
         # It may be useful for computing the terminal reward or debugging.
-        register_variables(
-            self.simulator.controller, self.controller.get_fieldnames(),
-            self._action, self.controller_name)
+        register_variables(self.simulator.controller,
+                           self.controller.get_fieldnames(),
+                           self._action,
+                           self.controller_name)
 
     def compute_command(self,
                         measure: SpaceDictNested,
@@ -534,9 +536,6 @@ def compute_command(self,
         :param measure: Observation of the environment.
         :param action: High-level target to achieve.
         """
-        # Backup the action
-        set_value(self._action, action)
-
         # Update the target to send to the subsequent block if necessary.
         # Note that `_observation` buffer has already been updated right before
         # calling this method by `_controller_handle`, so it can be used as
@@ -554,8 +553,9 @@ def compute_command(self,
         if self.simulator.is_simulation_running:
             # Do not update command during the first iteration because the
             # action is undefined at this point
+            set_value(self.env._action, self._target)
             set_value(self._command, self.env.compute_command(
-                self._observation, self._target))
+                self._observation, deepcopy(self._target)))
 
         return self._command