An accurate determination of muscle force is desired in many applications in different fields such as ergonomics, sports medicine, prosthetics, human-robot interaction and medical rehabilitation. Since individual muscle forces cannot be directly measured, force estimation using recorded electromyographic (EMG) signals has been extensively studied. This usually involves interpretation and analysis of the recorded EMG to estimate the underlying neuromuscular activity which is related to the force produced by the muscle. Although invasive needle electrode EMG recordings have provided substantial information about neuromuscular activity at the motor unit (MU) level, there is a risk of discomfort, injury and infection. Thus, non-invasive methods are preferred and surface EMG (SEMG) recording is widely used. However, physiological and non-physiological factors, including phase cancelation, tissue filtering, cross-talk from other muscles and non-optimal electrode placement, affect the accuracy of SEMG-based force estimation. In addition, the relative movement of the muscle bulk and the innervation zone (IZ) with respect to the electrode attached to the skin are two major challenges to overcome in force estimation during dynamic contractions. The objective of this work is to improve the accuracy of SEMG-based force estimation under static conditions, and devise methods that can be applied to force estimation under dynamic conditions. To achieve this objective, a novel calibration technique is proposed, which corrects for variations in the SEMG with changing joint angle. In addition, a modeling technique, namely parallel cascade identification (PCI) that can deal with non-linearities and dynamics in the SEMG-force relationship is applied to the force estimation problem. Finally, a novel integrated sensor that senses both SEMG and surface muscle pressure (SMP) is developed and the two signal modalities are used as input to a force prediction model. The experimental results show significant improvement in force prediction using data calibrated with the proposed calibration method, compared to using non-calibrated data. Joint angle dependency and the sensitivity to the location of the sensor in the SEMG-force relationship is reduced with calibration. The SEMG-force estimation error, averaged over all subjects, is reduced by 44\% for PCI modeling compared to another modeling technique (fast orthogonal search) applied to the same dataset. Significantly improved force estimation results are also achieved for dynamic contractions when joint angle based calibration and PCI are combined. Using SMP in addition to SEMG leads to significantly better force estimation compared to using only SEMG signals. The proposed methods have the potential to be combined and used to obtain better force estimation in more complicated dynamic contractions and for applications such as improved control of remote robotic systems or powered prosthetic limbs. / Thesis (Ph.D, Electrical & Computer Engineering) -- Queen's University, 2013-08-20 20:46:56.897

No description available.

No description available.

No description available.

Direct measurements of external forces acting on a structure are infeasible in many cases. The Augmented Kalman Filter (AKF) has several attractive features that can be utilized to solve the inverse problem of identifying applied forces, as it requires the dynamic model and the measured responses of structure at only a few locations. But, the AKF intrinsically suffers from numerical instabilities when accelerations, which are the most common response measurements in structural dynamics, are the only measured responses. Although displacement measurements can be used to overcome the instability issue, the absolute displacement measurements are challenging and expensive for full-scale dynamic structures. In this paper, a reliable model-based data fusion approach to reconstruct dynamic forces applied to structures using heterogeneous structural measurements (i.e., strains and accelerations) in combination with AKF is investigated. The way of incorporating multi-sensor measurements in the AKF is formulated. Then the formulation is implemented and validated through numerical examples considering possible uncertainties in numerical modeling and sensor measurement. A planar truss example was chosen to clearly explain the formulation, while the method and formulation are applicable to other structures as well.

<p>Most of the heavy-duty machines, in particular construction vehicles, employ hydraulically actuated functions that are used to perform multiple tasks with elevated power requirements. Such high-power demand motivates the Original Equipment Manufacturers (OEMs) to minimize the costs associated with energy consumption through the design of such hydraulic systems. The human-machine interaction (human factors) and the efficiency of the hydraulic control system are considered key elements towards a successful design. The interaction between the operator and the machine considerably affects the performance of construction machines. In order to maintain high levels of productivity, the operators require comfort and effortless controllability of the multiple hydraulic functions. The comfort requirement can include limited shocks and oscillations while operating the machines (while driving and controlling the implement motion), cabin accessories (AC, radio, cameras, etc.) and accessibility to the instrumentation. Besides, the operators have to control multiple functions simultaneously in an efficient manner while maintaining high levels of productivity. Consequently, the operators require smooth controllability of such functions. Such demand can largely affect the efficiency of the expected hydraulic control system and can induce additional costs and complexity. The OEMs are therefore forced to find a balance between efficiency and operators’ requirements to be competitive on the market. As a result, the currently adopted hydraulic architectures rely on purely hydraulic components to ensure robustness and functionality of the hydraulic functions at the expenses of limited performance and high-power consumption. In this dissertation, electro-hydraulic components are employed to induce improvements of the commercially available solutions while still complying with the operators’ demands and energy efficiency. To this end, this work tackles the weaknesses of traditional hydraulic architectures and it proposes alternative solutions to overcome their limitations. Two full-size wheel loaders are used to study the behavior of the existing system and later to implement the proposed variations. First, the development of an innovative ride control feature to improve the operators’ comfort is presented. Experimental results show the proposed strategy having better comfort performances compared to the purely hydraulic solution. Besides, the electro-hydraulic alternative does not demand the costly additional components the commercial solution instead requires. Second, this work faces the concern for efficiency of the present hydraulic architecture. The most diffused hydraulic system for the studied category of construction machines, commonly known as Load Sensing (LS), is sized to work most efficiently for elevated power conditions. During this work, an electronically controlled hydraulic supply unit and a flow-sharing method are used to reduce the hydraulic power consumption in the regions where the traditional LS system is less efficient. With a simple and cost-effective modification, the presented control strategy can induce an efficiency improvement over a wide range of operating conditions. Third, this dissertation proposes an operator-assistance feature to potentially increase the overall productivity and reduce the operator’s stress. An online estimation algorithm was developed to predict the payload weight of the transported material inside the bucket and the pushing forces during a typical loading cycle. The calculated payload mass provides an estimate of the user’s productivity level and it is extremely advantageous when the loaded material should reach a certain target weight. The developed estimation algorithm can also support an optimized autonomous excavation process, which can progressively limit the operator-machine interaction.</p>

The haptic device recreates the sense of touch by applying forces to the user. Since the device is “rendering” forces to emulate the physical interaction, the force control is essential for haptic devices. While a dedicated force/torque sensor can close the loop of force control, the additional equipment creates extra moving mass and inertia at the tool center point (TCP). Therefore, estimating the Cartesian force at the TCP has continuously been receiving attention over the past decades. The objective of this thesis project is to develop a real-time force estimation algorithm based on the proportional current-torque relationship with the dynamic modeling of the TAU haptic device. The algorithm can be further used for the force control of the device. The research questions of the thesis are: how to design and develop an algorithm for the TAU that used for Cartesian contact force estimation, how to set up the force estimation test bench and how to evaluate the results of the force estimation algorithm. In order to achieve the force estimation algorithm, a virtual environment is built to simulate the real-time haptic physics. Then an external force/torque sensor is installed at the TCP to get the measurement of the Cartesian force at the TCP. The force estimation algorithm calculates the Cartesian force at the TCP based on the current measurement of the DC motors at the six joints. The estimation result of the Cartesian force at the TCP is then compared with the force/torque sensor measurement to determine if the estimation algorithm is sufficiently accurate. The analysis of the estimation accuracy emphasizes the feasibility of Cartesian force estimation on the TAU haptic device. / En haptikenhet gör det möjligt att förmedla en känsla av kontakt i en virtuell värld genom att skapa krafter som motverkar en rörelse . Hur denna kraft skapas och kontrolleras är av stor vikt för att få den så verklighetstrogen som möjligt. Om man har en kraftsensor kan den användas till att utforma en kraftreglering med återkoppling, men på bekostnad av en ökad massa och tröghet vid användarens hand. Detta har medfört ett ökat intresse under de senaste åren för att på olika sätt försöka uppskatta den kraft som återkopplas till användaren utan att behöva en kraftsensor. Målet för detta examensarbete är att utveckla en algoritm för att uppskatta en kontaktkraft i realtid baserat på antagandet att motormomentet är proportionellt beroende av strömmen. Algoritmen kan sedan användas för att konstruera en sluten reglerloop med kraftåterkoppling för en haptisk enhet. Forskningsfrågorna som behandlas i detta examensarbete är;  hur kan vi utforma en algoritm för estimering av kontaktkrafter för haptikenheten TAU  hur kan vi utforma en experimentell försöksuppställning för mätning av de verkliga kontaktkrafterna från TAU vid kontakt.  hur kan vi använda resultaten från experimenten för utvärdering av algoritmen För testning och utvärdering av algoritmen har en virtuell värld skapats för att efterlikna en simuleringsmiljö som haptikenheten är tänkt att användas i. En kraftsensor har monterats under det verktyg som användaren håller i när enheten används när ett typiskt ingrepp ska övas i en simulator, t.ex. borrning i en tand. Vid experimenten beräknar algoritmen den uppskattade kontaktkraften som användaren känner baserat på den uppmätta strömmen för de sex motorer som aktiveras av kontakten. Dessa beräknade värden har sedan jämförts med de från kraftsensorn uppmätta för att avgör om algoritmen är tillräckligt noggrann. Analysen visar att noggrannheten är tillräckligt bra för att vara en lovande ansats till att användas för kraftuppskattning vid reglering av kontaktkraft för haptikenheten TAU.

No description available.

Patients suffering from brachial plexus injury or other spinal cord related injuries often lose their hand functionality. They need a device which can help them to perform day to day activities by restoring some form of functionality to their hands. A popular solution to this problem are robotic exoskeletons, mechanical devices that help in actuating the fingers of the patients, enabling them to grasp objects and perform other daily life activities. This thesis presents the design of a novel exoskeleton glove which is controlled by a neural network-based controller. The novel design of the glove consists of rigid double four-bar linkage mechanisms actuated through series elastic actuators (SEAs) by DC motors. It also contains a novel rotary series elastic actuator (RSEA) which uses a torsion spring to measure torque, passive abduction and adduction mechanisms, and an adjustable base. To make the exoskeleton glove grasp objects, it also needs to have a robust controller which can compute forces that needs to be applied through each finger to successfully grasp an object. The neural network is inspired from the way human hands can grasp a wide variety of objects with ease. Fingertip forces were recorded from a normal human grasping objects at different orientations. This data was used to train the neural network with a R2 value of 0.81. Once the grasp is initiated by the user, the neural network takes inputs like orientation, weight, and size of the object to estimate the force required in each of the five digits to grasp an object. These forces are then applied by the motors through the SEA and linkage mechanisms to successfully grasp an object autonomously. / Master of Science / Humans are one of the few species to have an opposable thumb which allows them to not only perform tasks which require power, but also tasks which require precision. However, unfortunately, thousands of people in the United States suffer from hand disabilities which hinder them in performing basic tasks. The RML glove v3 is a robotic exoskeleton glove which can help these patients in performing day to day activities like grasping semi-autonomously. The glove is lightweight and comfortable to use. The RML glove v3 uses a neural network based controller to predict the grasp force required to successfully grasp objects. After the user provides the required input, the glove estimates the object size and uses other inputs like object orientation and weight to estimate the grasp force in each finger linkage mechanism. The motors then drive the linkages till the required force is achieved on the fingertips and the grasp is completed.

Robots that support humans in physically demanding tasks require accurate force sensing capabilities. A common way to achieve this is by monitoring the interaction with the environment directly with dedicated force sensors. Major drawbacks of such special purpose sensors are the increased costs and the reduced payload of the robot platform. Instead, this thesis investigates how the functionality of such sensors can be approximated by utilizing force estimation approaches. Most of today’s robots are equipped with rich proprioceptive sensing capabilities where even a robotic arm, e.g., the UR5, provides access to more than hundred sensor readings. Following this trend, it is getting feasible to utilize a wide variety of sensors for force estimation purposes. Human proprioception allows estimating forces such as the weight of an object by prior experience about sensory-motor patterns. Applying a similar approach to robots enables them to learn from previous demonstrations without the need of dedicated force sensors. This thesis introduces Behavior-Specific Proprioception Models (BSPMs), a novel concept for enhancing robotic behavior with estimates of the expected proprioceptive feedback. A main methodological contribution is the operationalization of the BSPM approach using data-driven machine learning techniques. During a training phase, the behavior is continuously executed while recording proprioceptive sensor readings. The training data acquired from these demonstrations represents ground truth about behavior-specific sensory-motor experiences, i.e., the influence of performed actions and environmental conditions on the proprioceptive feedback. This data acquisition procedure does not require expert knowledge about the particular robot platform, e.g., kinematic chains or mass distribution, which is a major advantage over analytical approaches. The training data is then used to learn BSPMs, e.g. using lazy learning techniques or artificial neural networks. At runtime, the BSPMs provide estimates of the proprioceptive feedback that can be compared to actual sensations. The BSPM approach thus extends classical programming by demonstrations methods where only movement data is learned and enables robots to accurately estimate forces during behavior execution.