WalkMan

Project

WALK-MAN targets at enhancing the capabilities of existing humanoid robots, permitting them to assist or replace humans in emergency situations. These include rescue operations in damaged or dangerous sites like destroyed buildings or power plants. The robot will demonstrate :

  • dexterous, powerful manipulation skills - e.g. turning a heavy valve or lifting collapsed masonry
  • robust balanced locomotion - walking, crawling over a debris pile, etc.
  • physical sturdiness - e.g. operating conventional hand tools such as pneumatic drills or cutters.

The robot will show human levels of locomotion, balance and manipulation and operate outside the laboratory environment. The scenario challenges the robot in several ways: walking on unstructured terrain, in cluttered environments, among a crowd of people as well as crawling over a debris pile. The project's results will be evaluated using realistic scenarios, also consulting civil defense bodies.


UCL tasks

UCL leads the tasks relative to robot simulation and is highly involved in whole-body locomotion tasks.

Regarding simulation we aim at developing a symbolic model of the robot dynamics using the ROBOTRAN multibody software. An accurate simulator of the robot and its environment will be used for hardware design requirements and for fast software implementation and testing (Model in the loop). A real-time simulator will be used for tele-operation and to improve robot capabilities with embodied simulation (internal model).

Simulator Download

A complete version of simulator can be downloaded from gitlab:

https://gitlab.robotran.be/walkman/coman_robotran.git

This model includes 6 DoF for each leg, 3 DoF for waist and torso, and 7 DoF for each shoulder, arm and fore arm. The dynamics of the series elastic actuators and also ground reaction forces are modeled. In order to run this simulator, the user should have ROBOTRAN multibody software installed. The simulator is completely open source and any recommendations or suggestions are welcome.

 Locomotion

Regarding locomotion, we will design a dynamic walking gait controller based on a bio-inspired approach. The combination of three different kinds of walking “primitives”: feed-forward pattern generation mechanisms, short latency feedback reflexes, and long latency feedback postural control, will be exploredto optimize the energy exchanges between the robot and its environment. The purpose of this strategy is to steer the robot within hazardous and unstructured environments.

 

Collaboration

This project is conducted within the European Community's 7th Framework Programme through a collaboration between UCL and :

      

 

 

Contact

Renaud Ronsse

Paul Fisette

Nicolas Docquier

Nicolas Van der Noot

Timothée Habra

Alexandra Zobova

 

©2017 Robotran