Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


BinHammamPhDDissertation.pdf (5.97 MB)

ETD Abstract Container

Abstract Header

Whole-Body Motion Retargeting for Humanoids

Bin Hammam, Ghassan Mohammed
http://rave.ohiolink.edu/etdc/view?acc_num=osu1408367811

Abstract Details

2014, Doctor of Philosophy, Ohio State University, Electrical and Computer Engineering.
Humanoid motion generation using human-to-humanoid motion transfer (motion retargeting) has become an essential and useful method to produce human-like motions. However, this method has a number of challenges that limit its usefulness. The objective of this dissertation is to develop efficient methods to retarget whole-body, task-space motion from a human to a humanoid while managing online (real-time) dynamic (balance) and kinematic (joint-limit, self-collision, and foot) constraints, and while ensuring motion tracking characterized by dynamic consistency with natural human motion. This dissertation describes a very efficient, modified resolved acceleration control (MRAC) algorithm for dynamic filtering and control of whole-body humanoid motion in response to upper-body task specifications, or a commanded joint-space motion reference in general. The dynamic filter is applicable for general motions when standing in place. It is characterized by modification of the commanded torso acceleration based on a geometric solution to produce a Zero Moment Point (ZMP) which is inside the foot support. The resulting feasible, modified motion is synchronized to the reference motion when the computed ZMP for the reference motion again falls within the support. Contact forces at each foot are controlled through a dedicated force distribution module which optimizes the ankle roll and pitch torques. MRAC uses time-local information and is therefore targeted for online control. The effectiveness of the algorithm is demonstrated by means of simulated experiments. This dissertation presents a Cartesian-space constrained resolved acceleration control (CRAC) framework to manage execution of operational motion-tracking tasks, and handle constraints for redundant and non-redundant task specifications. The approach is particularly well suited for online control of humanoid robots using captured human motion data expressed by Cartesian variables. The current formulation is dynamically consistent, and enforces kinematic constraints such as joint-limit, self-collision, and foot constraints. Based on the applied reference motion and constraints setup, CRAC enables motion control at rates of 3k Hz in the fastest case and more than 400 Hz in the slowest case. A method, called the Unified CRAC (UCRAC), that handles dynamic balance and kinematic constraints together for whole-body task-space motion is also developed. The approach is non-iterative and as a result suitable for real-time applications. It utilizes an efficient centroidal dynamics formulation to relate the net force, applied to the humanoid by the environment, to the joint accelerations that realize the motion. UCRAC uses heuristically-designed rules to force the computed ZMP and centroidal projection to remain inside the foot support area, and to compute and command the net force on the system. Enforcing foot constraints are managed through the application of the constrained centroidal dynamic equations. UCRAC is capable to handle quasi-static tic balance in addition to dynamic balance, which is shown to provide significant flexibility for different humanoid foot dimensions. This method performs with rates providing faster than real-time performance with a minimum speed of 250 Hz which includes all constraint computations as well. The efficacy of the proposed methods is demonstrated in all cases by simulated and real-time experiments of task-level human motion replication on a Honda-like humanoid robot model.
David Orin, Prof. (Advisor)
Yuan Zheng, Prof. (Committee Member)
Kevin Passino, Prof. (Committee Member)
172 p.
Computer Engineering; Electrical Engineering; Robotics
Humanoid; Motion retargeting; centroidal dynamics; human-like motions; dynamic constraints; balance; kinematic constraints; joint-limit; self-collision; Cartesian-space controller; resolved acceleration control; dynamically consistent

Recommended Citations

Citations

  • Bin Hammam, G. M. (2014). Whole-Body Motion Retargeting for Humanoids [Doctoral dissertation, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu1408367811

    APA Style (7th edition)

  • Bin Hammam, Ghassan. Whole-Body Motion Retargeting for Humanoids. 2014. Ohio State University, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=osu1408367811.

    MLA Style (8th edition)

  • Bin Hammam, Ghassan. "Whole-Body Motion Retargeting for Humanoids." Doctoral dissertation, Ohio State University, 2014. http://rave.ohiolink.edu/etdc/view?acc_num=osu1408367811

    Chicago Manual of Style (17th edition)

osu1408367811
893
© 2014, all rights reserved.
This open access ETD is published by The Ohio State University and OhioLINK.