Kinova GEN3 experiment using Vector Field strategy in SE(3)

Steps for reproducing the experiment.

Kinova API

Kortex API v2.6.0 was used

# Create conda environment for Python 3.9
conda create -n kinova python=3.9
conda activate kinova

# Install Kinova API
git clone https://github.com/verlab/demos-verlab.github
cd demos-verlab
git checkout c079f4e
cd kinova-demos
pip install -e .

Uaibot

Unpublished private uaibot version was used. Requires Python 3.11.

git clone git@github.com:fbartelt/uaibot_experimental.git
cd uaibot_experimental
git checkout 5750bd6

python3.11 -m venv py311
source py311/bin/activate
pip install .

Setup

Kinova robot uses TCP for communication. Configure the network:

chmod +x kinova_network_setup.sh 
./kinova_network_setup.sh

To revert network changes:

chmod +x reset_network.sh
./reset_network.sh

Run

The experiment relies on two Python environments. Both scripts share data using numpy array shared memory

Precompute curve in SE(3)

Generate and resample the curve for uniform distribution in SE(3). The result is saved in resampled_curve.npy

source py311/bin/activate
py311/bin/python ./precompute_curve.py

Control

Run the control and experiment scripts (order doesn't matter):

source py311/bin/activate
py311/bin/python ./control.py

conda activate kinova
python ./experiment.py

Visualization

control.py stores experiment data in data.pkl. Access it with:

import pickle

with open("./data.pkl", "rb") as f:
    data = pickle.load(f)

config_hist = data["config_hist"]
hist_index = data["hist_index"]
hist_dist = data["hist_dist"]

Visualize results using check_experimental_results.py in the py311 environment.

Animation

experiment.py stores configurations and timestamps in config_data.pkl. Access it with:

import pickle

with open('config_data.pkl', 'rb') as f:
    data = pickle.load(f)

config_hist = data['config_hist']
time_hist = data['time_hist']

Animations can be visualized by running the notebook cells in experiment_animation.py using VS Code's IPython interactive cells or by converting the file into a Jupyter notebook.