add pretrained checkpoints, update tiago model files

Author: dHonerkamp

.dockerignore

@@ -6,6 +6,5 @@ scripts/wandb/
 scripts/paper_n1/
 scripts/paper_n2/
 webots/
-scripts/model_checkpoints/
 .git/objects
 *.npy

Dockerfile

@@ -55,7 +55,6 @@ RUN catkin config --blacklist tiago_pcl_tutorial # combined_robot_hw_tests # for
 RUN /bin/bash -c ". /opt/ros/melodic/setup.bash && catkin build || exit"
 
 # update the moveit configs with the global joint
-COPY gazebo_world/tiago/robot/modified_tiago.srdf.em src/tiago_moveit_config/config/srdf/tiago.srdf.em
 COPY gazebo_world/tiago/robot/modified_tiago_pal-gripper.srdf src/tiago_moveit_config/config/srdf/tiago_pal-gripper.srdf
 COPY gazebo_world/tiago/robot/modified_gripper.urdf.xacro src/pal_gripper/pal_gripper_description/urdf/gripper.urdf.xacro
 COPY gazebo_world/tiago/robot/modified_wsg_gripper.urdf.xacro src/pal_wsg_gripper/pal_wsg_gripper_description/urdf/gripper.urdf.xacro

gazebo_world/tiago/robot/modified_tiago.srdf.em

@@ -1,20 +1,16 @@
-<!-- Autogenerated file, don't edit this, edit tiago.srdf.em instead-->
 <?xml version="1.0" ?>
+<!-- =================================================================================== -->
+<!-- |    This document was autogenerated by xacro from /dev/stdin                     | -->
+<!-- |    EDITING THIS FILE BY HAND IS NOT RECOMMENDED                                 | -->
+<!-- =================================================================================== -->
 <!--This does not replace URDF, and is not an extension of URDF.
     This is a format for representing semantic information about the robot structure.
     A URDF file must exist for this robot as well, where the joints and the links that are referenced are defined
 -->
 <robot name="tiago">
-    <!--GROUPS: Representation of a set of joints and links. This can be useful for specifying DOF to plan for, defining arms, end effectors, etc-->
-    <!--LINKS: When a link is specified, the parent joint of that link (if it exists) is automatically included-->
-    <!--JOINTS: When a joint is specified, the child link of that joint (which will always exist) is automatically included-->
-    <!--CHAINS: When a chain is specified, all the links along the chain (including endpoints) are included in the group. Additionally, all the joints that are parents to included links are also included. This means that joints along the chain and the parent joint of the base link are included in the group-->
-    <!--SUBGROUPS: Groups can also be formed by referencing to already defined group names-->
     <group name="base">
         <joint name="world_joint" />
     </group>
-
-
     <group name="arm">
         <joint name="arm_1_joint" />
         <joint name="arm_2_joint" />
@@ -48,7 +44,7 @@
       <group name="base" />
    </group>
     <!--END EFFECTOR: Purpose: Represent information about an end effector.-->
-    <end_effector name="gripper" parent_link="arm_tool_link" group="gripper" parent_group="arm_torso" />
+   <end_effector group="gripper" name="gripper" parent_group="arm_torso" parent_link="arm_tool_link" />
 
  <virtual_joint name="world_joint" type="planar" parent_frame="odom" child_link="base_footprint" />
     <!--PASSIVE JOINT: Purpose: this element is used to mark joints that are not actuated-->

gazebo_world/tiago/robot/modified_tiago_pal-gripper.srdf

@@ -37,7 +37,7 @@
       <xacro:wrong_wrist_model/>
     </xacro:if>
   </xacro:if>
-  <xacro:if value="${end_effector not in ['false', False, 'pal-gripper', 'pal-hey5', 'schunk-wsg', 'robotiq-2f-85', 'robotiq-2f-140', 'custom']}">
+  <xacro:if value="${end_effector not in ['false', False, 'pal-gripper', 'pal-hey5', 'schunk-wsg', 'robotiq-2f-85', 'robotiq-2f-140', 'robotiq-epick', 'custom', 'no-ee']}">
     <xacro:wrong_end_effector/>
   </xacro:if>
   <xacro:if value="${ft_sensor not in ['false', False, 'schunk-ft']}">
@@ -57,6 +57,7 @@
    <xacro:property name="has_arm"                 value="$(arg arm)"/>
    <xacro:property name="has_schunk_ft"           value="${ft_sensor == 'schunk-ft'}"/>
    <xacro:property name="has_ft_sensor"           value="${has_schunk_ft}"/> <!-- This should be OR'ed with other FT sensors-->
+   <xacro:property name="has_end_effector"        value="${end_effector not in ['false', False, 'no-ee']}"/>
    <xacro:property name="end_effector_link"       value="${'wrist_ft_tool_link' if has_ft_sensor else 'arm_tool_link'}"/>
    <xacro:property name="end_effector_name"       value="${'hand' if end_effector == 'pal-hey5' else 'gripper'}"/>
    <xacro:property name="head_link_name"          value="head_2_link"/>
@@ -107,7 +108,7 @@
 	  <xacro:tiago_arm name="arm" parent="torso_lift_link" wrist_model="${wrist_model}" has_ft_sensor="${has_ft_sensor}" reflect="1" tiago_dual="-1" arm_1_offset="${arm_1_joint_offset}" arm_2_offset="${arm_2_joint_offset}" arm_3_offset="${arm_3_joint_offset}" arm_4_offset="${arm_4_joint_offset}" arm_5_offset="${arm_5_joint_offset}" arm_6_offset="${arm_6_joint_offset}" arm_7_offset="${arm_7_joint_offset}" no_safety_eps="${no_safety_eps}">
         <origin xyz="0.15505 0.014 -0.151" rpy="0 0 ${-90 * deg_to_rad}" />
       </xacro:tiago_arm>
-      <xacro:tiago_end_effector name="${end_effector_name}" parent="${end_effector_link}" type="${end_effector}" reflect="1"/>
+      <xacro:tiago_end_effector name="${end_effector_name}" parent="${end_effector_link}" type="${end_effector}" has_end_effector="${has_end_effector}" reflect="1"/>
   </xacro:if>
 
   <!-- Force Torque sensor -->

gazebo_world/tiago/robot/tiago.urdf.xacro

@@ -4,7 +4,7 @@ Repository providing the source code for the paper "N2M2:  Learning Navigation f
 Please cite the paper as follows:
 
     @article{honerkamp2021learning,
-        title={Learning Navigation for Arbitrary Mobile Manipulation Motions in Unseen and Dynamic Environments},
+        title={N2M2: Learning Navigation for Arbitrary Mobile Manipulation Motions in Unseen and Dynamic Environments},
         author={Daniel Honerkamp and Tim Welschehold and Abhinav Valada},
         journal={arXiv preprint arXiv:2206.08737},
         year={2022},
@@ -38,6 +38,14 @@ The following commands also assume that you have docker installed, and to use a
 
 The commands to train and evaluate the models in the paper can be found in `docker_run.sh`.
 
+To run the container interactively, use the following commands: 
+
+    docker run -it --gpus all --rm dhonerkamp/mobile_rl:11.1-latest bash
+    cd catkin_ws_[pr2/hsr/tiago]
+    source devel/setup.bash
+    conda activate
+
+then proceed to the "Run" section below.
 
 ## Evaluating on your own task
 Tasks are implemented as environment wrappers in `scripts/modulation/envs/tasks.py`.
@@ -68,8 +76,8 @@ Install moveit and the pr2
     sudo apt-get install ros-[version]-pr2-simulator
     sudo apt-get install ros-[version]-moveit
     sudo apt-get install ros-[version]-moveit-pr2
-    ros-[version]-pr2-common
-    ros-[version]-pr2-navigation
+    ros-melodic-pr2-common
+    ros-melodic-pr2-navigation
 
 Create a catkin workspace (ideally a separate one for each robot)
 
@@ -78,7 +86,7 @@ Create a catkin workspace (ideally a separate one for each robot)
 
 Install further ros packages:
 
-    yes | rosinstall src /opt/ros/[ros-version] modulation.rosinstall
+    yes | rosinstall src /opt/ros/melodic modulation.rosinstall
 
 Fork the repo and clone into `./src`
 
@@ -110,15 +118,21 @@ For more details and how to install the ROS requirements for the other robots pl
 
 
 ### Run
+- If you have a weights and biases account, log into the account. Otherwise first run `wandb offline` in your shell, which will turn off logging.
+
 - Training (set workers etc. as needed, this is the command used in the paper):
 
         python src/modulation_rl/scripts/main_ray.py --load_best_defaults --num_gpus 0.24 --num_workers 20 
         --num_cpus_per_worker 0.1 --num_envs_per_worker 1 --num_gpus_per_worker 0 --ray_verbosity 2
         --training_intensity 0.1 --load_best_defaults --vis_env
 
-- Evaluation:
+- Evaluation of a pretrained checkpoint (pick one of [pr2/hsr/tiago]):
+
+        python src/modulation_rl/scripts/evaluation_ray.py --evaluation_only --eval_execs sim gazebo --env [pr2/hsr/tiago] --model_file=[pr2/hsr/tiago]/checkpoint-1000 --eval_tasks picknplace --vis_env 
+
+- Evaluation of a wandb logged run (adjust [pr2/hsr/tiago] and ${wandb_run_id}):
 
-        python src/modulation_rl/scripts/evaluation_ray.py --evaluation_only --eval_execs sim gazebo --env pr2 --resume_id ${wandb_run_id} --eval_tasks picknplace --vis_env
+        python src/modulation_rl/scripts/evaluation_ray.py --evaluation_only --eval_execs sim gazebo --env [pr2/hsr/tiago] --resume_id ${wandb_run_id} --eval_tasks picknplace --vis_env
 
 5. [Only to visualise] start rviz: