Printer Friendly

Modular control for a six-legged walking machine based on the joint posture.

Abstract: Starting point in the presented paper is the building of a basic module for control of the leg's joint followed by setting the control architecture through coupling such modules. Objective of the approach is to achieve a timely synchronized control/motioning of the leg using pneumatic actuators. Key words: walking robot, fluidic muscles, modular control


One disadvantage of fluidic muscles when set to drive a joint is that unlike electrical motors they cannot create rotary motions but rather perform linear translation (shortening and extension). The assignment to control a joint consists in the pressure and force control of the contracting muscle, respectively of the joint's rotation angle. As far as the air passing through the muscle provides for the sole control loop variable the first step should be to generate a basic scheme for controlling the joint. Next comes the conception enhancing, i. e. the definition of elementary behavior types featured by each joint along with verifying the way this types need to be combined in order to settle complex control architecture regarding the leg.


A joint is viewed as the smallest functional unit. The task hereby means to control the actuator in a way that achieves the necessary rotation angle for a given joint. Since fluidic muscles generate only one-direction forces a joint actuated by a pair of muscles--protagonist/flexor and antagonist/extensor was put into use. The next level functional unit is the leg composed of three joints--h, v and k (Fig. 1.). The goal is to accomplish a timely synchronized rotation of the joints, each in a preset angle. Such synchronization is the essential precondition for meeting the requirement towards maintaining constant body height H and foot stretch S.

Considering the way they are featured throughout specific writings close-loop control conceptions dealing with antagonistic fluidic-muscle-based actuators can be referred to as either "bang-bang" or "impulse control" (Berns & Grimminger et al., 2003). The recited methods and experiments were exercised using three posture valves equipped with springs for reverting to a starting position after completing the relevant operation. The control valve serves both muscles--the flexor [U.sub.ext] and the extensor [U.sub.fex]--and is specified by the state of the variable in equation E. Equation 1 shows the rightward rotation scheme; in case of a leftward rotation both muscles exchange roles.

[U.sub.fex] = |E [right arrow 1]/E [right arrow 0] [U.sub.fex] = |E [right arrow -1]/E [right arrow 0] (1)

where: E = 0--valve is closed, E = 1--valve lets air into the flexor, E = -1--valve lets air out of the extensor. In the case of a bang bang control the variable E keeps a stable value until reaching the deliberated posture/rotation. This control mode is very quick but inclined to fluctuations. To enhance the control over air flows passing through the muscles actuator and to avoid any fluctuation trends a "pulsating valve" as described in (Weidemann&Pfeiffer et al., 2004) can be applied to the effect of a very short-time-opening and immediate closing of the latter. In this case E translates as "pulsation frequency" (a minus frequency value denotes the airflow leaving the actuator) (Kerscher&Albiez et al., 2002). The duration of the impulse sequence depends on the variable E.


The classical approach towards a robot control involves the determination of ingredient trajectories by way of a timely synchronization of kinematics, parametrizng of body motions and legs trajectories. Such control architectures had been brought about foremost by electrical drives but their direct application for controlling pneumatic actuators proves to be rather complicated. The reason behind the fact is that the latter cannot be controlled in terms of speed. A simulation executed on MatLab and regarding a pneumatically actuated six-legged mobile robot with a step d=23 cm provided value changes for angles h, v and k in all three leg joints (Fig, 2 and Fig. 3). A timely non-synchronization of the occurring angles. A simulation executed on MatLab and regarding a pneumatically actuated six-legged mobile robot with a step d=23 cm provided value changes in angles h, v and k for all three leg joints (Fig. 2 and 3). A timely non-synchronization of the occurring angles presets trajectory deviations or changes in the body posture even in flat floor facilities. A non-synchronization of the angles provokes trajectory deviations or changes in the body posture even on flat floor.

This paper assumes that synchronization could be achieved through implying premeditated checkpoints where the slower joints are waited for and than all three joints--concurrently started anew. To define the checkpoints several conditions have to be considered:

* Start and end positions of the joints

* "Zero" values for the angles

* Time delays when angles do not or change insufficiently

* Direction change in the joint rotation.

For the case outlined in this paper the points could be set as: Initial value-[T.sub.s] for the joint h=-0,5 rad, for v=-0,01 rad and for k=-0,135 rad; regarding the "End" value [T.sub.e]--for h=0,5 rad, for v=0,01 rad and for k=0,135 rad. "Zero" values for the angles and change in the rotation direction--To. It would be purposeful to introduce similar checkpoints regarding a linear interval v preconditioned by T1 for h=-0,3 rad, for v=-0,001 rad and for k=-0,04 rad and T2 for h=0,3 rad, for v=0,001 rad and for k=0,04 rad.



The whole logic design presented in this paper is hierarchical i.e. the top-level design file is VHDL based and includes component instances, that refer to schematic based ones. The register transfer level architecture of the joint block follows the logic shown on Fig. 4. The main units in the design are an 8-bits digital comparator and two RS Flip-Flops. The comparator provides relation between the predefined externally set up value and the current position of the joint achieved through an external sensor also with 8 bits of resolution. Equality of two values activates an output of the comparator, which is used in conjunction with two external signals to act as reset for the flip-flops. To set purposes for the logic blocks a 2 bits external signal for motion control is also put into use. Outputs of the flip-flops are accordingly used for enabling external actors to induce left or right rotation movements of the joint as can be read from the figure. Included is a general enabling signal for the digital comparator, which empowers the entire joint block. Hence, up till the current position value of the joint reaches the predefined position value the actors for rotation and linear movements remain active.




For development purposes the integrated synthesis environment (ISE) WebPACK was used, which is the ideal solution for the FPGA and CPLD design. (Cohen, 1999). Finally, the simulation of the described system using a third-party simulator from Cadence--ModelSIM (Fig. 5) was examined. The result proves the generation of control signals for a two-direction-rotating as was mentioned earlier.




The presented paper offers a structure intended at significant enhancement in the modularity of control for mobile robots, which use the principle of peristaltic motion through antagonistically functioning muscles. Mechanisms intended hereby to perform a synchronized control of the joints can also be applied for improving the foot's trajectory and hence lead to essential increase in stability. This will also grant a wider scope of implementation of these particular robots. The exercised impulse valve control secures a pulse length of 48 ms with max frequency of 20 Hz. In further researches the announced control architecture will be enriched by additional level and control mechanisms with reflex combinations on diverse levels.


Every research this paper is accounting for has been done under the assignment of Contract BY--TH--201/2006 entitled "Research of a Modular Architecture for the Control of Mechatronic Elastic Multi-Link Devices"


Berns K.; Grimminger, F., Hochholdinger, U., Kerscher, T. & Albiez, J. (2003) Design and Control of a Leg for the Running Machine PANTER. Proceedings of the 11th International Conference on Advanced Robotics, 2003, p.p. 1737-1742

Kerscher T.; Albiez, J. & Berns, K. (2002) Joint control of the six-legged robot AirBug driven by fluidic muscles. Proceedings of the Third International Workshop on Robot Motion and Control, Poland, 2002.

Weidemann, H.-J.; Pfeiffer, F. & Eltze, J. (2004) The six-legged TUM walking robot. Intelligent Robots and Systems (IROS), Volume 2, 2004, p.p. 1026-1033.

Cohen, B. (1999). VHDL coding styles and methodology, Kluwer Academic Publishers, ISBN 0-7923-8474-1, Boston, MA, USA
COPYRIGHT 2007 DAAAM International Vienna
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2007 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Milushev, Mladen; Georgieva, Vania
Publication:Annals of DAAAM & Proceedings
Article Type:Report
Geographic Code:1USA
Date:Jan 1, 2007
Previous Article:Direct implementation of control algorithms from simulation environment into PLC.
Next Article:Using cutting tools durability in optimizing concentrated fabrication system flow simulation.

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |