M. Stelzer

Towards Bipedal Jogging as a Natural Result of Optimizing Walking Speed for Passively Compliant Three-Segmented Legs

Elasticity in conventionally built walking robots is an undesired side-effect that is suppressed as much as possible because it makes control very hard and thus complex control algorithms must be used. The human motion apparatus, in contrast, shows a very high degree of flexibility with sufficient stability. In this research we investigate how compliance and damping can deliberately be used in humanoid robots to improve walking capabilities. A modular robot system consisting of rigid segments, joint modules and adjustable compliant cables spanning one or two joints is used to configure a human-like biped. In parallel, a simulation model of the robot was developed and analyzed. Walking motion is gained by oscillatory out-of-phase excitations of the hip joints. An optimization of the walking speed has been performed by improving the viscoelastic properties of the leg and identifying the appropriate hip control parameters. A good match was found between real robot experiments and numerical simulations. At higher speeds, transitions from walking to running are found in both the simulation as well as in the robot.

Efficient Walking Speed Optimization of a Humanoid Robot

The development of optimized motions of humanoid robots that guarantee fast and also stable walking is an important task, especially in the context of autonomous soccer-playing robots in RoboCup. We present a walking motion optimization approach for the humanoid robot prototype HR18 which is equipped with a low-dimensional parameterized walking trajectory generator, joint motor controller and an internal stabilization. The robot is included as hardware-in-the-loop to define a low-dimensional black-box optimization problem. In contrast to previously performed walking optimization approaches, we apply a sequential surrogate optimization approach using stochastic approximation of the underlying objective function and sequential quadratic programming to search for a fast and stable walking motion. This is done under the conditions that only a small number of physical walking experiments should have to be carried out during the online optimization process. For the identified walking motion for the considered 55 cm tall humanoid robot, we measured a forward walking speed of more than 30 cm s -1 . With a modified version of the robot, even more than 40 cm s -1 could be achieved in permanent operation.

Biologically Inspired Reflex Based Stabilization Control of a Humanoid Robot with Artificial SMA Muscles

Suddenly occurring collisions or unintentional motions represent a high safety risk in robotics and must be prevented. Especially for humanoid robots, the influence of disturbances that occur unexpectedly during bipedal locomotion are difficult to compensate. A model based online control approach for stabilization of a humanoid robot with many degrees of freedom may require too much time for computing and implementing an adequate compensating motion. In addition, such a control approach usually requires accurate sensor information about the type and magnitude of the disturbance. The goal of the present paper is a reflex based online stabilization control of a humanoid robot actuator based on artificial SMA muscles. The design of a humanoid robot actuated with SMA muscles allows a lightweight robot design and simplifies the direct implementation of reflexes. The reflex that is integrated into the robot depends on an evaluation of the pressure distribution of the feet. An instable position of the center of mass of the robot leads to a known specific pressure disturbance that should be avoided. The experiments show that the implementation of a reflex for the actuators in the calf leads to a stabilization of the entire robot. Additional reflexes are required when the strength or speed of disturbances are increased, such as in the upper leg or arms.

MACROSCOPIC SMA WIRE BUNDLE ACTUATOR/SENSOR SYSTEM: DESIGN, MEASUREMENT, CONTROL APPROACH

Abstract Prestrainend shape memory alloys (SMA) change their length when heated above their transformation temperature. Based on this property this paper presents the design of macroscopic SMA actuators scalable in force and length that keeps up a short cool down time to guarantee a high frequency of contraction/stress cycles and the possibility of arranging the fixings in any direction. A new model of the macroscopic actuator has been developed. The model describes the actuators behaviour and offers the possibility to use the resistance of the actuator as a linear position encoder. Experimental results demonstrate that the newly developed SMA device can be used as actuator and position sensor. The measurement shows that the fixings of the actuator can be shifted or rotated without influence on the actuators behaviour and therefore various uses are possible. Based on the measurement a first control approach has been developed and tested.

Walking, Running and Kicking of Humanoid Robots and Humans

In this paper key aspects and several methods for modeling, simulation, optimization and control of the locomotion of humanoid robots and humans are discussed. Similarities and differences between walking and running of humanoid robots and humans are outlined. They represent several, different steps towards the ultimate goals of understanding and predicting human motion by validated simulation models and of developing humanoid robots with human like performance in walking and running. Numerical and experimental results are presented for model-based optimal control as well as for hardware-in-the-loop optimization of humanoid robot walking and for forward dynamics simulation and optimization of a human kicking motion.

Control approach for a novel high power-to-weight robotic actuator scalable in force and length

The development of a control approach for a novel, soundless, lightweight and multifunctional shape memory alloy (SMA) actuator scalable in force and length for personal assistance or home-help robots is presented in this paper. The SMA actuator is based on lightweight bundles of thin wires of prestrained shape memory alloy that change their length when heated above their transformation temperature. The design approach of the actuator allows arranging the point of actuation in any direction and ensures a short cool down time to guarantee a frequency of contraction/stress cycles that is high enough to allow fast joint motions. This is needed for the generation of fast joint motions. For the use of the actuator the novel control approach has been experimentally validated. The approach uses the resistance of the actuator as a linear position encoder and there are no additional external sensors needed. The application of the new actuator to a novel lightweight humanoid robot is outlined. One advantage of the actuator over electric motors lies in the large variety of user-defined points of actuation of the in pull-force and length free scalable actuators and the high power-to-weight ratio. The results demonstrate that it is possible to build a large humanoid robot actuated with SMA actuator in a new way.

Human Kicking Motion Using Efficient Forward Dynamics Simulation and Optimization

The problem of finding and predicting muscle activations for free goal oriented or measured human motion is one of the basic problems in biomechanics. While currently inverse dynamics approaches are most commonly used for their computational efficiency, they can not handle the problem in its most general form. We here present computational methods that increase the computational efficiency of the forward dynamics approach by two orders of magnitude. Results are presented for a time optimal kicking motion and the analysis of a measured kicking motion. Current work includes investigation of a jumping motion and finding optimal walking motions for a robot that is driven by artificial muscles.

Laufbewegungen bei Roboter, Tier und Mensch : Analyse, Modellierung, Simulation und Optimierung

Die Wurzeln der Forschung zum Verstandnis der komplizierten Laufprozesse bei Robotern, Tieren und Menschen liegen in der Biologie, der Medizin und den Ingenieurwissenschaften. In der Biologie und Medizin ist das Interesse u.a. durch die Diagnose und bessere Heilung von Laufproblemen und die Entwicklung intelligenter, aktiver Beinprothesen begrundet. Die Ingenieurwissenschaften sind z.B. an Bau und Betrieb von Laufmaschinen zur Fortbewegung in unebenem Gelande und der Integration von interner und externer Sensorik und Motorik in die zielorientierte Planung und Steuerung der Laufbewegung autonomer Roboter interessiert. Fur alle diese Untersuchungen sind Modellierung und Simulation der Dynamik des Laufens von zentraler Bedeutung.