Department for Automation, Biocybernetics and Robotics





Biologically inspired synthesis of periodic movement for a robotic humanoid leg

Project duration: 2010 - 2012
Project area: Automation and Intelligent Control of Robots
Project type: Research
Project funding: Research Agency
Project leader: Andrej Gams

Abstract

Control of the humanoid robot lower extremities is a complex problem and one, which has not been satisfactorily solved for tasks as bipedal walking or vertical jump. Despite the complexity of the mentioned tasks, humans can perform them with ease. Performing such task with robots demands new, alternative approaches, inspired on biological systems.

Project description

Control of the humanoid robot lower extremities is a complex problem and one, which has not been satisfactorily solved for tasks as bipedal walking or vertical jump. Despite the complexity of the mentioned tasks, humans can perform them with ease. Performing such task with robots demands new, alternative approaches, inspired on biological systems.

Bilogical systems perform various aperiodic and periodic movements, which we can effectively imitate with robots. The first steps in imitating have been made also in the field of imitating the synthesis of movement of biological systems, where appropriate motion can be synthesized using chains of nonlinear oscillators. We will use such an approach for controlling a humanoid robotic leg, which can be perceived of as a planar multi-segmented mechanism. The basic hypothesis of the proposed research is, whether we can use the method of synthesizing movement, based on biological systems, to control a humanoid robotic leg in a vertical plane, and whether we can modify the method to primary demands of such task with an emphasis on stability.

The basic idea is based on imitating biological systems and their neurological signals with nonlinear oscillators. The final goal is the development and implementation f periodic movement of a humanoid robot leg with a new method for synthesizing periodic movements, and combining the movement with aperiodic motion – the vertical jump. The secondary goal is the upgrading the experimental mechanism into a bipedal mechanism.

Idea origin: The idea of controlling a humanoid robot extremity derives from a method of controlling multi-segmental planar robotic mechanism that model a robotic snake or robotic salamander. Besides the similarity in the structure of a robotic snake and a robotic leg (despite the difference in number and length of the segments), we originate also from the possibility of using nonlinear oscillators for synthesis of periodic movement of separate segments, the characteristics of nonlinear oscillator synchronization properties and the use of pairs of oscillators to simulate virtual antagonistically acting muscles.

Problem: The main problem of the research is in including the necessary condition og maintaining stability in a vertical plain. Also demanding and complex is the mechanical structure of the mechanism. It has been shown that a human leg has biarticular tendons that allow better energy efficiency in both walking and vertical jump. The task demands modification of an existing mechanism to include a biarticular tendon that acts only when the knee is being extended. This implies a nonlinear mechanical coupling, which is superior to the mathematical coupling of some of the oscillators and joints. As such, the task demands novel approaches for classifying their interactions.

Goals: The main goal is the development and application of periodic movement of a robotic humanoid lower extremity mechanism with the use of a chain of nonlinear oscillators. The control must ensure the stability of the robot. The synthesis of the movement must produce more human-like movement than approaches of model-based control. The system must account for the mechanical coupling of the biarticular tendon and achieve energy efficient movement, as it has to ensure also the control in the dynamical phases of the movement. The secondary goal includes expanding the robotic mechanism to a bipedal mechanism and performing dynamic walking.

Originality: Nonlinear oscillator based methods have never been applied on a humanoid robot mechanism in the described fashion. Available mechanisms very rarely include biarticular tendons, which is essential for efficient walking and successful vertical jumping. The combination of both can turn out to be the crucial break-through in the field of dynamic movement designing for humanoid robots.

Methods: The proposed research demands measurements of human movement using a system for motion capture, measurements of the acting forces and signals in the muscles. Using mathematical modeling we can ensure appropriate behavior of virtual and mechanical systems. The mechanical system demands an upgrade to include the biarticular tendon.

Relevance: The proposed research is based on research published in journals with highest impact factors in their fields. It is in some way similar, but not identical to the EU project EVRYON, from the EU FP7 package, which proposes chains of nonlinear oscillators for control of robotic exoskeletons for humans.

Publications

Article

  • Petrič T., Gams A., Ijspeert A., Žlajpah L., On-line frequency adaptation and movement imitation for rhythmic robotic tasks, 2011. [More]

Inproceedings

  • Gams A., Petrič T., Babič J., Ude A., Žlajpah L., Robotsko posnemanje gibanja s prilagajanjem spremenjivi podlagi, 2011. [More] [PDF]
  • Gams A., Petrič T., Babič J., Žlajpah L., Posnemanje gibanja z ohranjanjem stabilnosti : robotski počep = Movement imitation with maintaining stability : robot squatting, 2011. [More]
  • Gams A., Petrič T., Babič J., Žlajpah L., Ude A., Constraining movement imitation with reflexive behavior : robot squatting, 2011. [More]
  • Gams A., Petrič T., Babič J., Žlajpah L., Ude A., Constrained mapping of movement from human to robot : robot sqatting, 2011. [More]
  • Gams A., Petrič T., Babič J., Žlajpah L., Augmenting movement imitation with reflexive stability behavior, 2011. [More]