MORMG | Multi-Objective Reactive Motion Generation in Mobile-Manipulators

Multi-objective reactive motion generation based on composable energy policy

TIAGo++ mobile manipulator

Installation

• Tested in Python 3.7

Install the repository. Inside the repository

pip install -e .

Install all packages

pip install -r requirements.txt

Requirements

pip install numpy matplotlib pybullet future

• Pinocchio

conda install pinocchio -c conda-forge

• PyTorch

conda install pytorch torchvision torchaudio cudatoolkit=10.2 -c pytorch

Run test for OSC(Operation Space control)

• Test simple OSC only for left-arm control

python scripts/darias_energy_control/OSC/TIAGo_OnePoint_Baseline.py

• Test simple OSC for whole-body control(Left arm and base)

python scripts/darias_energy_control/OSC/TIAGo_OnePoint_with_holoBaseCXY.py

Run test for CEP(Composable Energy Control)

• Test simple original CEP by running the python file

python scripts/darias_energy_control/test_simple_darias.py

python scripts/darias_energy_control/test_simple_tiago.py

• Test CEP of TIAGo++ with different energy components

python scripts/darias_energy_control/_test_tiago_lefthand_base_taskgotoAndpathplan.py

python scripts/darias_energy_control/_test_cep_tiago_lefthand_base_pathplan.py

python scripts/darias_energy_control/_test_simple_tiago_lefthand_base.py

Result pictures

• Cascade control

• CEP bayesian network for TaskGoto and PathPlan energy components

• RRT tree

• Vector fields based on rrt tree

• One path generated by rrt-based vector field

• Heatmap for closest point

TODO list