RTL-planning

This repo contains the code base for the below project, which is accepted to appear in IROS 2024.

Paper

Farhad Nawaz, Shaoting Peng, Lars Lindemann, Nadia Figueroa, Nikolai Matni, "Reactive Temporal Logic-based Planning and Control for Interactive Robotic Tasks", arXiv preprint arXiv:2404.19594, 2024. (available at https://arxiv.org/abs/2308.00186).

TL;DR: We build a discrete task planner (200 Hz) and a continuous motion planner (1 KHz) for reactive time-critical tasks and complex periodic motions. Project webpage: https://sites.google.com/view/lfd-neural-ode/home

Dataset

The Data Demos folder contains the trajectory data for all the experiments. We train on this data to learn the Neural ODE models.

Automaton graph and python simulation

The $\texttt{Notebooks}$ folder contains the following two jupyter notebooks.

1. automaton_graph.ipynb: to generate the automaton graph given a Reactive Temporal Logic (RTL) specification using the Spot 2.0 tool. It also has classes to compute the shortest path on the graph and choose controllable propositions, given the uncontrollable propositions (environmental observations).
2. STL_ My_wiping_nODE.ipynb: Code to create toy python simulations that illustrate a periodic 2D wiping task with a reaching motion specified by an RTL specification. The reaching motion is executed using a time-varying Control Barrying Function (CBF).

Implement on the Franka robot arm

Follow the installation instructions from our prior work to setup and build the ROS workspace for both gazebo simulation and real robot implementation on the Franka robot arm.

Experiment setup

• Move the files from the config and launch folder to /franka_interactive_controllers/config/ and /franka_interactive_controllers/launch/, respectively. The config folder containts the Neural ODE models (.eqx), reference trajectories (.npy) and automaton graphs (<graph.dictionary>) for planning.

• For a specific , move all the scripts from <path-to-scripts> to /franka_interactive_controllers/scripts/, where <path-to-scripts>:=/Gazebo_scripts/<experiment> for gazebo simulation, or /Lab_PC_scripts/<experiment> for real robot implementation.

• For gazebo, check the launch file /franka_interactive_controllers/launch/simulate_panda_gazebo.launch for the initial joint configuration of the task you want to run.

Running the experiment

• In the first terminal, run

roslaunch franka_interactive_controllers <launch-file> controller:=passiveDS_impedance

where, launch-file:=simulate_panda_gazebo.launch for gazebo simulation, or franka_interactive_bringup.launch for real robot implementation. We can also change the controller from passiveDS_impedance to cartesian_twist_impedance, etc.

• In the second terminal, run

roslaunch franka_interactive_controllers NODE_model.launch

Change the name of the ROS nodes to <node>_SO3.py in the launch file for cooking experiment.

• In the third terminal, run

rosrun franka_interactive_controllers cmd_vel.py

Run cmd_vel_SO3.py instead for cooking experiment.