SLMC //student projects: Biped Humanoid Locomotion//

SLMC homeSLMC researchSLMC peopleSLMC projectsSLMC newsSLMC coursesSLMC hiringSLMC contactIPABSchool of InformaticsUniversity of Edinburgh
Project Leader
Yariv Bachar
MSc, School of Informatics, University of Edinburgh
Project Supervisor
Sethu Vijayakumar, PhD
IPAB, School of Informatics, Univ. of Edinburgh

Project Title

Developing Controllers for Biped Humanoid Locomotion

Slope

Project Goal

For decades biped humanoid locomotion has been a major focus of robotics research. Many theoretical studies along with much simulation and practically realized systems have been researched up to the Honda and Sony humanoid robots - the most modern and advanced biped locomotion robots.

Maintaining dynamic stability during locomotion is fundamental for developing realized humanoid robots which are able to operate in real world environment.

This project involves with generating walking gaits for a full body humanoid simulation with physics (already available through Webots simulation software). The simulator already provides the basic commands to sense and command individual motors. The key challenge is to devise a control scheme which will ensure robust walking pattern and the ability to recover from stumbling.

The principle goal of the project is to develop a natural looking walking gait (initially static, later dynamic) for a full body 25DOF humanoid robot. Ideally, our aim was to incorporate Zero Moment Point (ZMP) control for generating dynamically stabled walking.

Programming interface is in C/C++ and there are sample modules available (in Webots) which assist in tracing individual joint trajectories.

The project is decomposed to the following phases:

  1. Literature review and Environment setup for simulations:
    1. Literature review of related research - material collection and reading of related papers.

    2. Webots and additional libraries and packages setup.

  2. Design and implementation of ZMP control algorithms/methods:
    1. ZMP statically stable control implementation.
    2. ZMP dynamically stable control implementation.
  3. Experimenting with the various ZMP control methods and adjustment:
    1. Running simulations using various control methods implemented.
    2. Evaluation and analysis of results.
  4. Write-up/Conclusion:
    1. Preparing the dissertation.
    2. Presentation of the project.

Project Timeline

Time Frame Task (completed or scheduled)  
1 May. 04 - 21 May. 04 Material collection and reading; Environment setup. C
22 May. 04 - 26 Jun. 04 Control algorithms/methods design and implementation. C
10 Jun. 04 Group Review meetings. C
27 Jun. 04 - 27 Jul. 04 Experimenting with control algorithms/methods; Improvement of control methods. C
28 Jul. 04 - 1 Sep. 04 Dissertation write-up; Preparing project presentation. C
2 Sep. 04 Dissertation submission. C
3 Sep. 04 - 10 Sep. 04 Project presentation. C

Project Results & Conclusions

MSc Thesis (PDF, 4.23 MB).

Presentation slides (ZIP, 2.68 MB).

Source Code (TGZ, 392 KB).

Snapshot I - Humanoid walking in a downhill experiment.

Snapshot II - Humanoid walking with external force experiment.

Snapshot III - The Humanoid in a slope environment.

Movie I - The humanoid walking on the standard horizontal stage.

Movie II - The humanoid walking on the standard environment with external head wind force.

Links

Robotics Research

Tools

  • Webots - Award winning mobile robotics simulation software.

  • Roboop - A robotics object oriented package in C++.

  • NEWMAT - A comprehensive matrix library in C++.

  • Arabica - An XML parser toolkit, providing SAX2 and DOM implementations.

  • ODE - Open Dynamics Engine by Russell Smith.

  • CGAL - Computational Geometry Algorithms Library in C++.

  • OpenGL - The Industry's Foundation for High Performance Graphics.

  • OSE - a generic application framework for constructing general purpose applications.