The HUMANOID POSTURAL DYNAMICS project (2006 – present)

Acronym: POSTURAL DYNAMICS - II
Name: Humanoid Postural Dynamics
Type: Collaborative project
Funds: French Ministery of Research - CNRS - Univ. Montpellier-1 - Univ Montpellier-2 - French-Japan Joint Robotics Laboratory (JRL)
UM1 key researchers: Benoît Bardy, Julien Lagarde, Sofiane Ramdani
Collaborators: Philippe Fraisse (PI), Vincent Bonnet, Nacim Ramdani, Philippe Poignet, (Dept Robotics, LIRMM-Univ Montpellier 2)

HUMANOID POSTURAL DYNAMICS is a theoretical and experimental project aiming at bio-inspiring the command and control of postural coordination in humanoid robots based on concepts and data obtained in humans (see the POSTURAL DYNAMICS - I project ). Due to their anthropomorphic structure, humanoid robots often present dynamic similarities with humans, and the basic knowledge of coordination principles can be helpful to design parsimonious, flexible yet stable humanoid postural control commands. Vice versa, the successful implementation of new postural algorithms in humanoids may orient research in human postural control in new directions, and may be used to predict the behavior of the postural system under specific circumstances (e.g., its stability limits under constraints such as age, obesity, pathology, etc...). 

Theory and current research:

The repeated findings obtained in human posture research that postural modes (i) emerge out of the coalescence of multiple constraints, (ii) exhibit persistences and changes that are characteristic of self-organized systems (e.g., Bardy et al., 1999, 2002) have oriented recent research in humanoid robotics. In this new research program, our aim is (i) to build a simple but realistic biomechanical model of the human body that exhibit the hallmarks observed in humans, and (2) to implement the observed solutions in two humanoid robots, HOAP3 and HRP2, under appropriate intial conditions and under-constraints optimization principles. The tracking task tested by Bardy et al. (1999) is used again here for this purpose. The recent results (e.g., Bonnet et al., 2007, 2008, 2009a, 2009b) evidence the reproduction of several hallmarks observed in humans (stable in phase and antiphase modes, transition betwen them) but not others (hysteresis, critical fluctuations). Both HOAP3 and HRP2 spontaneously reproduce these patterns when asked to maintain the center of mass above the feet while moving the head with appropriate amplitude and frequency.

Click here to see HOAP3 performing an anti-phase pattern

Key references (downloadable version in page Vitae):

1. Bonnet, V., Lagarde, J., Fraisse, P., Ramdani, N., Ramdani, S., Poignet, P., Bardy, B. G. (2007) Modeling human postural coordination to improve the control of balance in humanoids. IEEE International Conference on Humanoid Robots, Humanoids 2007.
2. Bonnet, V., Fraisse, P., Ramdani, N., Lagarde, J., Ramdani, S., & Bardy, B. G. (2008). Modeling postural coordination dynamics using a closed-loop controller. IEEE International Conference on Humanoid Robots, Humanoids 2008.
3. Bonnet, V., Fraisse, P., Ramdani, N., Lagarde, J., Ramdani, S., & Bardy, B. G. (2009a). A robotic closed-loop scheme to model human postural coordination. In Proceedings of the 2009 IEEE/RSJ International Conference on Intelligent RObots and Systems (IROS-2009).
4. Bonnet, V., Fraisse, P., Ramdani, N., Lagarde, J., Ramdani, S., & Bardy, B. G. (2009b). A closed loop musculoskeletal model of postural coordination dynamics. In Proceedings of the 48th IEEE-Conference on Decision and Control.
5. Bonnet, V., Ramdani, S., Fraisse, P., Ramdani, N., Lagarde, J., & Bardy, B. G. (2011). A structurally optimal control model for predicting and analyzing human postural coordination. Journal of Biomechanics, in press.