#slam #robotics #math

gs-rs

Implementation of a basic monocular SLAM (Simultaneous Localization and Mapping) Algorithm in pure Rust

1 unstable release

0.1.0 Aug 31, 2020

#316 in Math

MIT/Apache

2.5MB
2K SLoC

gs-rs [ˈdʒiːzrs] - GraphSLAM in Rust

Travis CI Coverage Status License-MIT License-Apache Issues Forks Stars Crates.io Version

gs-rs is an implementation of a monocular graph based SLAM backend in pure rust. It is largely inspired by the General Graph Optimizer g2o of Rainer Kümmerle, though it is giving it a rusty touch.

Overview

Documentation

A brief introduction to the algorithms used is available as a Jupyter Notebook here

Building Instructions

  • Install the rust toolchain
  • Clone the repository
  • Execute cargo build --release in the root directory

Example Usage

In addition to the code snippet below, further examples for different use cases can be found in examples/. Information regarding that directory can be found in its respective README file.

use gs_rs::parser::json::JsonParser;
use gs_rs::parser::Parser;
use gs_rs::optimizer::optimize;
use gs_rs::visualizer::visualize;

fn main() {

    // parse file at "file_path.json" to internal factor graph representation
    let factor_graph = JsonParser::parse_file("file_path.json").unwrap();

    // optimize the factor graph's variables with 5 iterations
    optimize(&factor_graph, 5);

    // display the improved factor graph in a new window
    visualize(&factor_graph);

}

Attribution

The file format .g2o as well as the reference implementation in C++ are are part of the project g2o.

The datasets used in data_files/ and examples/ have been described by Carlone et al. and adapted to include fixed vertices. Other modifications have been performed as well to demonstrate the different vertex and edge types supported in gs-rs. The original datasets as well as more examples can be found online. Attribution and details may also be found in the data sets section.

Note

We, the contributers are fully aware of the fact that we cannot prohibit military usage of open source software. But we want to state very clearly that we discourage any use of our software with the intention of harming people.

License

© 2020 Samuel Valenzuela, TNG Technology Consulting GmbH

© 2020 Florian Rohm, TNG Technology Consulting GmbH

© 2020 Daniel Pape, TNG Technology Consulting GmbH

gs-rs is licensed under the Apache License, Version 2.0 (cf. here or http://www.apache.org/licenses/LICENSE-2.0) or the MIT license (cf. here or http://opensource.org/licenses/MIT), at your option.

Dependencies

~14MB
~263K SLoC