Eric Kubica

Introduction of the Foot Placement Estimator: A Dynamic Measure of Balance for Bipedal Robotics

The goal of most bipedal robotics research is to develop methods of achieving a dynamically balanced gait. Most current approaches focus on maintaining the balance of the system. This paper introduces a measure called the foot placement estimator (FPE) to restore balance to an unbalanced system. We begin by developing a theoretical proof to define when a biped is stable, as well as defining the region in which stability results are valid. This forms the basis for the derivation of the FPE. The results of the FPE are then extended to a complete gait cycle using the combination of a state machine and simple linear controllers. This control system is applied to a detailed and realistic simulation based on a physical robot currently under construction. Utilizing the FPE as a measure of balance allows us to create dynamically balanced gait cycles in the presence of external disturbances, including gait initiation and termination, without any precalculated trajectories.

Perception-based lossy haptic compression considerations for velocity-based interactions

The ability of technology to transmit multi-media is very dependent on compression techniques. In particular lossy compression has been used in image compression (jpeg) audio compression (mp3) and video compression (mpg) to allow the transmission of audio and video over broadband network connections. Recently the sense of touch or haptics is becoming more important with its addition in computer games or in cruder applications such as vibrations in a cell phone. As haptic technology improves the ability to transmit compressed force sensations becomes more critical. Most lossy audio and visual compression techniques rely on the lack of sensitivity in humans to pick up detailed information in certain scenarios. Similarly limitations in the sensitivity of human touch could be exploited to create haptic models with much less detail and thus requiring smaller bandwidth. The focus of this paper is on the force thresholds of the human haptic system that can be used in a psychophysically motivated lossy haptic (force) compression technique. Most of the research in this field has measured the just noticeable difference (JND) of the human haptic system with a human user in static interaction with a stationary rigid object. In this paper our focus involves cases where the human user or the object are in relative motion. An example of such an application would be the haptic rendering of the user’s hand in contact with of a high-viscous material or interacting with a highly deformable object. Thus an approach is presented to measure the force threshold based on the velocity of the user’s hand motion. Two experiments are conducted to detect the absolute force threshold (AFT) of the human haptic system using methodologies from the field of psychophysics. The AFTs are detected for three different ranges of velocity of the user’s hand motion. This study implies that when a user’s hand is in motion fewer haptic details are required to be stored calculated or transmitted. Finally the implications of this study on a more complete future study will be discussed.

Stabilization of a Dynamic Walking Gait Simulation

Forward dynamic simulations of human walking gait have typically simulated and analyzed a single step of the walking cycle, assuming symmetric and periodic gait. To enable simulations over many steps, a stabilizer is required to maintain the balance of the walking model, ideally mimicking the human balance control mechanism. This paper presents a feedback control system that stabilizes the torso orientation during a human walking gait dynamic simulation, enabling arbitrarily long simulations. The model is a two-dimensional mechanical simulation, in which the desired joint trajectories are defined as functions of time; the only external forces on the model are gravitational and ground reaction forces. Orientation or postural control is achieved by modulation of the rate at which lower limb joints move through angular trajectories. The controller design is based on a sequence of simple linear feedback controllers, each based on an intuitive control law. Controller parameters were determined iteratively using an optimization algorithm and repeated executions of the forward dynamics simulation to minimize control term errors. Results show the use of feedback control and joint speed modulation to be effective in maintaining balance for walking simulations of arbitrary length, allowing for analysis of steady-state walking.

Multi-Step Forward Dynamic Gait Simulation

Forward dynamic simulations of anthropomorphic human models are becoming increasingly investigated as an alternative to inverse dynamic analyses ([1] and [2]). Forward dynamic models are typically forced into a walking pattern using pre-computed joint trajectories reminiscent of a central pattern generator. Simple balance controllers inspired from similar work in the robotics field [3] allow the models to walk for many steps. The current work implements a 2D, 7-segment, 9-degree-of-freedom gait model coupled with an optimization routine to find a minimally fatiguing gait.

A fuzzy control strategy for a flexible single link robot

A fuzzy control strategy is described which is utilized to control the rigid body and the first flexural mode of vibration in a single link robotic arm. Both simulated and experimental results are presented and show that the rigid and flexible modes can be adequately controlled with the proposed technique. The results are shown to provide some improvement over those obtained by more conventional means.<<ETX>>

Designing stable MIMO fuzzy controllers

This paper presents a systematic procedure for constructing a multi-input multi-output fuzzy controller that guarantees identical performance to an existing stabilizing linear controller. An algorithm is devised that generates a fuzzy controller which is functionally identical to a given time-invariant or time-varying finite-dimensional linear controller. The benefit of this transformation is that it provides an automated technique for the initial fuzzy controller setup while vital knowledge-based attributes are integrated afterwards. An important result of this work is that once a linguistic mapping into the fuzzy domain has been performed, one can see in linguistic terms how the linear controller operates. The effectiveness of this approach is demonstrated with a model for a flexible robot that exhibits nonminimum phase characteristics. An extension is outlined to use this deterministic approach for the case of a general dynamic control-law, and several applications to nonlinear control problems are discussed.

Perception and Generation of Affective Hand Movements

Perception and generation of affective movements are essential for achieving the expressivity required for a fully engaging human-machine interaction. This paper develops a computational model for recognizing and generating affective hand movements for display on anthropomorphic and non-anthropomorphic structures. First, time-series features of these movements are aligned and converted to fixed-length vectors using piece-wise linear re-sampling. Next, a feature transformation best capable of discriminating between the affective movements is obtained using functional principal component analysis (FPCA). The resulting low-dimensional feature transformation is used for classification and regeneration. A dataset consisting of one movement type, closing and opening the hand, is considered for this study. Three different expressions, sadness, happiness and anger, were conveyed by a demonstrator through the same general movement. The performance of the developed model is evaluated objectively using leave-one-out cross validation and subjectively through a user study, where participants evaluated the regenerated affective movements as well as the original affective movements reproduced both on a human-like model and a non-anthropomorphic structure. The proposed approach achieves zero leave-one-out cross validation errors, on both the training and testing sets. No significant difference is observed between participants’ evaluation of the regenerated movements as compared to the original movement, which confirms successful regeneration of the affective movement. Furthermore, a significant effect of structure on the perception of affective movements is observed.

A novel fuzzy framework for nonlinear system control

Nonlinear PID and gain scheduling controls have attracted much research interests in recent years. These control strategies can accommodate some nonlinear characteristics by allowing the gains varied or rescheduled as a function of system states. In this paper, a novel fuzzy framework is developed to transfer this type of nonlinear controller to its fuzzy domain representation. It is proved that the resulting fuzzy controller is functionally exactly identical to the original control system. One of the benefits of the suggested approach is that it provides a well-defined prototype for the design of fuzzy control system. The resulting linguistic representation can facilitate investigation in linguistic terms into how the controller operates, whereas expert knowledge can be effectively implemented to improve control performance. The viability of the proposed mapping technique is demonstrated by using simulations corresponding to a flexible-link robot.

Feedforward and deterministic fuzzy control of balance and posture during human gait

The primary objective of this research is to model the biomechanical control system employed by the central nervous system (CNS) to maintain posture and balance of the head-arms-torso (HAT) during gait. More specifically, the intent is to stabilize a model of the upper body so that the HAT response is similar to that found experimentally in human subjects during gait. This is accomplished by using appropriate physiological parameters as feedback to generate realistic control signals at the hip musculature. The modelling includes the HAT musculoskeletal characteristics as well as pure neural time delays. The CNS control system is modelled by a linear state feedback controller as well as a hybrid fuzzy controller that has been adapted from a linear quadratic regulator. In addition, a newly derived feedforward component is demonstrated. It is expected that a better understanding of the human fait process will prove valuable in designing assistive devices and bipedal robots.

Human Factors for Designing a Haptic Interface for Interaction with a Virtual Environment

Designing haptic displays is one of the main challenges in creating virtual reality systems with the sense of touch. The design of the hardware and software of haptic interfaces depends critically on the capabilities of the human haptic system. For example, force feedback interfaces, due to inherent hardware limitations such as friction and actuator saturation, present forces to users in the case of interactions with a virtual environment which are only approximations of the forces that they would feel if they were interacting with the real world. Thus, quantitative human studies are required to obtain the impact on human performance of these approximations in the forces from the haptic devices. First, the focus of this paper is on the quantitative measures of human factors (force thresholds) that affect the design specifications of force feedback haptic interfaces when the human user or the object are in relative motion. Second, the effects of two direction of forces and the increment/decrement of forces are also studied through two experiments. It appears that the JNDs of human force perception depend on the force direction and the force increment/decrement, and these two variables must be incorporated in an efficient haptic design technique when the users hand is in motion.