Tested in Matlab R2021a and R2021b
Written by Joohwan Seo, Ph.D. Student in Mechanical Engineering, UC Berkeley.
These are quite raw, unorganized files. I hope everyone can get some of the insights, and please forgive me for the nasty codes.
- There was a serious error in the benchmark controller; now that it is fixed, it works far better than before.
- For the readers to test controllers in various scenarios, more scenarios are added.
- In
desired_trajectory2.mandtrajectory_calculator.m, we show a way to design a polynomial-based smooth trajectory generation technique also for the rotational part.
- Final version of the paper submitted.
- Control equations are added on every main_geo_discrete file.
- Geometric Impedance Control version 2 added
- Additional updates on the trajectory tracking simulations.
- Code variables now match with the variables in the paper.
The arxiv paper is updated to the final version.
main_geo_discrete.m runs the simulation for the proposed approach (intuitive geometric impedance) - Equation (30)
main_geo_discrete_v2.m runs the simulation for the proposed approach (geometric impedance control-version 1, GIC-v1) - Equation (32)
main_geo_discrete_v3.m runs the simulation for the proposed approach (geometric impedance control-version 2, GIC-v2) - Equation (47)
main_imp_discrete.m runs the simulation for the benchmark approach (conventional impedance control, CIC)
main_imp_discrete_v2.m runs the simulation for the benchmark approach, version 2 (Not rigorously tested)
plotter.m for visualizing the result -- GIC-v1 vs CIC
plotter2.m for visualizing the result -- GIC-v2 vs CIC\
Comparison results on plotter2.m are not presented in the paper since the two control laws are different, thus unable to do a fair comparison.
plotter_geo_comp.m is for comparison between intuitive geometric impedance control and GIC-v1. When the gains become different, they start to behave differently.
When gains are scalar in translational and rotational ways, both controllers are identical.
Change obj variable to tracking, tracking2, regulation, and regulation2.
tracking: sinusoidal trajectory tracking (presented in paper)
tracking2: 3rd-order smooth polynomial trajectory tracking both in translational and rotational
regulation: regulation task
regulation2: Multi-point regulation, can be considered as step-input case.
desired_trajectory.m gives the trajectory utilized in tracking objective.
desired_trajectory2.m gives the trajectory utilized in tracking2 objective.
desired_trajectory_regulation.m gives the desired setpoint utilized in regulation objective.
desired_trajectory_regulation2.m gives the desired setpoints utilized in regulation2 objective.
RTB matlab (Robotics Toobox Matlab) is needed to run the code.
Build the UR5e robot model and associated dynamic parameter matrices as well as Jacobian matrices.
Dynamic parameter matrices built from function_generator are saved here. Some miscellaneous functions are also defined here.
The simulation result data generated by the main files are saved here.
IFAC World Congress 2023, Yokohama, Japan: "Geometric Impedance Control on SE(3) for Robotic Manipulators"
Arxiv submitted version: https://doi.org/10.48550/arXiv.2211.07945
Official Proceedings version: https://doi.org/10.1016/j.ifacol.2023.10.1581
@article{seo2023geometric,
title={Geometric impedance control on SE (3) for robotic manipulators},
author={Seo, Joohwan and Prakash, Nikhil Potu Surya and Rose, Alexander and Choi, Jongeun and Horowitz, Roberto},
journal={IFAC-PapersOnLine},
volume={56},
number={2},
pages={276--283},
year={2023},
publisher={Elsevier}
}
To be appeared and presented at American Control Conference (ACC) 2024, Toronto, Canada
"A Comparison Between Lie Group- and Lie Algebra- Based Potential Functions for Geometric Impedance Control"
https://arxiv.org/abs/2401.13190
https://github.com/Joohwan-Seo/Geometric_Impedance_Control_Comparison