Task-Specific Synthesis and Design of a mobile 6-DoF Hexa Parallel Robot for Weed Control

Authors: Tim Sterneck, Jannik Fettin, and Moritz Schappler

This code is supplemental material to the following publication at the I4SDG Workshop 2023:

@InProceedings{SterneckFetSch2023,
  author       = {Sterneck, Tim and Fettin, Jannik and Schappler, Moritz },
  booktitle    = {Proceedings of the 2nd IFToMM for Sustainable Development Goals Workshop},
  title        = {Task-Specific Synthesis and Design of a mobile 6-DoF Hexa Parallel Robot for Weed Control},
  note         = {submitted for publication},
  organization = {Springer},
  date         = {2023},
}

Contents and Useage

This repository contains Matlab scripts and Latex code to reproduce results and figures of the paper. Beforehand, the steps from the prerequesites section have to be performed.

1. To reproduce the results, first the dimensional synthesis has to be performed by running the Matlab script dimsynth/config_amun.m.
   • If existing results shall be evaluated, their location has to be set with i4sdg2023_dimsynth_data_dir.m. Then this step can be omitted
   • The results that were used in the paper are stored in data/synthesis_results and are used by default
2. The synthesis' results have to be post-processed by the scripts from dimsynth_postprocess (in this order):
   • eval_figures_pareto.m: Assemble all Pareto fronts for the individual robots
   • robot_names.m: The names of the robots are assembled (e.g. for the figure captions); only once.
   • eval_existing_design.m uses the synthesis output and computes the performance of the actual designed prototype described in the text of Sec. 4 of the paper
   • eval_figures_pareto_groups.m: Group the robots to a smaller set.
   • select_eval_robot_examples.m: Select the specific robot structures from the Pareto front for a detailed view
3. The robot figures and detail information is reproduced with the Matlab scripts from paper/figures:
   • robot_images.m: Creates one image file for each robot. These are taken in robots.svg to create Fig. 2 the paper. Run robots/config_robot_figs.m for additional formatting.
   • The Pareto diagram of Fig. 3,a of the paper is already formatted by eval_figures_pareto.m
   • Two redundancy map of Fig. 3,b of the paper is formatted by perfmap_fig.m.
   • The workspace figures of Fig. 5 are created by a script which is not ready for publication yet.

The steps 2 to 3 can be performed after each iteration of the dimensional synthesis by using run_evaluation_i4sdg2023.m in the paper repo's root directory.

Prerequisites

For the Matlab scripts to work, the following steps have to be performed:

1. Install Matlab (tested with R2022b)
2. Set up the Matlab mex compiler, if not already done
3. Set up the path to the results for this paper by copying i4sdg2023_dimsynth_data_dir.m.template without the template suffix to the same location and by changing the content of the file to point to the stored results of the synthesis.
4. Download Matlab dependencies:
   • Toolbox for geometry functions (tested with rev. c4af53d; 2021-10-26)
   • Toolbox for trajectory functions (tested with rev. 42c8302; 2022-02-09))
   • Collection of external dependencies (tested with rev. d8d29ff; 2022-08-26))
   • Toolbox for various Matlab functions (tested with rev. 0654bba; 2022-04-05))
   • Robotics toolbox (tested with rev. 24e9a94; 2023-01-29))
   • Serial robot model database (tested with rev. 36ab3b167; 2023-01-20))
   • Parallel robot model database (tested with rev. 7c31c32; 2023-01-24))
   • Toolbox for symbolic robot model generation (tested with rev. fd37713; 2022-01-20)
   • Toolbox for transferring jobs to a computing cluster (tested with rev. a4bb277; 2022-12-07))
   • Dimensional synthesis toolbox (tested with rev. 9c24493; 2023-01-29))
   • Update the repositories to the given versions. You should first try the more recent version before downgrading to the given versions. It should still work.
   • Short command to download all repositories (run in Git Bash in Windows or in Linux terminal):

     git clone [email protected]:SchapplM/robotics-dep-geometry.git
     git clone [email protected]:SchapplM/robotics-dep-traj.git
     git clone [email protected]:SchapplM/robotics-dep-ext.git -b dev
     git clone [email protected]:SchapplM/matlab_toolbox.git -b dev
     git clone [email protected]:SchapplM/robotics-toolbox.git -b dev
     git clone [email protected]:SchapplM/robsynth-serroblib.git -b dev
     git clone [email protected]:SchapplM/robsynth-parroblib.git -b dev
     git clone [email protected]:SchapplM/robsynth-modelgen.git
     git clone [email protected]:SchapplM/matlab_pbs_transfer.git
     git clone [email protected]:SchapplM/robsynth-structdimsynth.git -b dev

5. Set up the toolboxes by modifying the following files according to the instructions in the respective README files
   • robsynth-serroblib/maplerepo_path.m.template (has to link to robsynth-modelgen)
   • robsynth-structdimsynth/computingcluster_repo_path.m.template (has to link to matlab_pbs_transfer, if a cluster is used for computation)
6. Run the path initialization scripts (..._path_init.m) of all downloaded software repos in Matlab. Put the run commands in the startup.m file of Matlab if used frequently.