Structure learning of graphical models for task-oriented robot grasping

In the collective imaginaries a robot is a human like machine as any androids in science fiction. However the type of robots that you will encounter most frequently are machinery that do work that is too dangerous, boring or onerous. Most of the robots in the world are of this type. They can be found in auto, medical, manufacturing and space industries. Therefore a robot is a system that contains sensors, control systems, manipulators, power supplies and software all working together to perform a task. The development and use of such a system is an active area of research and one of the main problems is the development of interaction skills with the surrounding environment, which include the ability to grasp objects. To perform this task the robot needs to sense the environment and acquire the object informations, physical attributes that may influence a grasp. Humans can solve this grasping problem easily due to their past experiences, that is why many researchers are approaching it from a machine learning perspective finding grasp of an object using information of already known objects. But humans can select the best grasp amongst a vast repertoire not only considering the physical attributes of the object to grasp but even to obtain a certain effect. This is why in our case the study in the area of robot manipulation is focused on grasping and integrating symbolic tasks with data gained through sensors. The learning model is based on Bayesian Network to encode the statistical dependencies between the data collected by the sensors and the symbolic task. This data representation has several advantages. It allows to take into account the uncertainty of the real world, allowing to deal with sensor noise, encodes notion of causality and provides an unified network for learning. Since the network is actually implemented and based on the human expert knowledge, it is very interesting to implement an automated method to learn the structure as in the future more tasks and object features can be introduced and a complex network design based only on human expert knowledge can become unreliable. Since structure learning algorithms presents some weaknesses, the goal of this thesis is to analyze real data used in the network modeled by the human expert, implement a feasible structure learning approach and compare the results with the network designed by the expert in order to possibly enhance it.

Graphical Model-Based Vertebra Identification from X-Ray Image(s)

Automated identification of vertebrae from X-ray image(s) is an important step for various medical image computing tasks such as 2D/3D rigid and non-rigid registration. In this chapter we present a graphical model-based solution for automated vertebra identification from X-ray image(s). Our solution does not ask for a training process using training data and has the capability to automatically determine the number of vertebrae visible in the image(s). This is achieved by combining a graphical model-based maximum a posterior probability (MAP) estimate with a mean-shift based clustering. Experiments conducted on simulated X-ray images as well as on a low-dose low quality X-ray spinal image of a scoliotic patient verified its performance.

A review on probabilistic graphical models in evolutionary computation

Thanks to their inherent properties, probabilistic graphical models are one of the prime candidates for machine learning and decision making tasks especially in uncertain domains. Their capabilities, like representation, inference and learning, if used effectively, can greatly help to build intelligent systems that are able to act accordingly in different problem domains. Evolutionary algorithms is one such discipline that has employed probabilistic graphical models to improve the search for optimal solutions in complex problems. This paper shows how probabilistic graphical models have been used in evolutionary algorithms to improve their performance in solving complex problems. Specifically, we give a survey of probabilistic model building-based evolutionary algorithms, called estimation of distribution algorithms, and compare different methods for probabilistic modeling in these algorithms.

Selective activation of the rectus abdominis muscle during low-intensity and fatiguing tasks

In order to understand the potential selective activation of the rectus abdominis muscle, we conducted two experiments. In the first, subjects performed two controlled isometric exercises: the curl up (supine trunk raise) and the leg raise (supine bent leg raise) at low intensity (in which only a few motor units are recruited). In the second experiment, subjects performed the same exercises, but they were required to maintain a certain force level in order to induce fatigue. We recorded the electromyographic (EMG) activities of the lower and upper portions of the rectus abdominis muscle during the exercises and used spatial-temporal and frequency analyses to describe muscle activation patterns. At low-intensity contractions, the ratio between the EMG intensities of the upper and lower portions during the curl up exercise was significantly larger than during the leg raise exercise (p = 0.02). A cross-correlation analysis indicated that the signals of the abdominal portions were related to each other and this relation did not differ between the tasks (p = 0.12). In the fatiguing condition, fatigue for the upper portion was higher than for the lower portion during the curl up exercise (p = 0.008). We conclude that different exercises evoked, to a certain degree, individualized activation of each part of the rectus abdominis muscle, but different portions of the rectus abdominis muscle contributed to the same task, acting like a functional unit. These results corroborate the relevance of varying exercise to modify activation patterns of the rectus abdominis muscle.

Feature selection environment for genomic applications

Background: Feature selection is a pattern recognition approach to choose important variables according to some criteria in order to distinguish or explain certain phenomena (i.e., for dimensionality reduction). There are many genomic and proteomic applications that rely on feature selection to answer questions such as selecting signature genes which are informative about some biological state, e. g., normal tissues and several types of cancer; or inferring a prediction network among elements such as genes, proteins and external stimuli. In these applications, a recurrent problem is the lack of samples to perform an adequate estimate of the joint probabilities between element states. A myriad of feature selection algorithms and criterion functions have been proposed, although it is difficult to point the best solution for each application. Results: The intent of this work is to provide an open-source multiplataform graphical environment for bioinformatics problems, which supports many feature selection algorithms, criterion functions and graphic visualization tools such as scatterplots, parallel coordinates and graphs. A feature selection approach for growing genetic networks from seed genes ( targets or predictors) is also implemented in the system. Conclusion: The proposed feature selection environment allows data analysis using several algorithms, criterion functions and graphic visualization tools. Our experiments have shown the software effectiveness in two distinct types of biological problems. Besides, the environment can be used in different pattern recognition applications, although the main concern regards bioinformatics tasks.

Males still have limb asymmetries in multijoint movement tasks more than 2 years following anterior cruciate ligament reconstruction

More than 2 years after undergoing anterior cruciate ligament (ACL) reconstruction, women still present bilateral asymmetries during multijoint movement tasks. Given the well-known ACL-injury gender bias, the goal of this study was to investigate whether males also present such asymmetries more than 2 years after undergoing ACL reconstruction. This study involved 12 participants submitted to ACL reconstruction in the ACL group and 17 healthy participants in the control group. The mean postoperative period was 37 months. The participants executed bilateral countermovement jumps and load squat tasks. The kinematics and ground reaction forces on each lower limb and pelvis were recorded, and used to compute bilateral peak vertical ground reaction forces, peak knee and hip joint powers in the sagittal plane, and the ratio between these powers. For the jump task, the groups had the same performance in the jump height, but for the ACL group the peak knee joint power on the operated side was 13% lower than on the non-operated side (p = 0.02). For the squat task, the hip-knee joint power ratio on the operated side of the ACL group was 31% greater than on the non-operated side (p = 0.02). The ACL group presented a deficit in the operated knee that had its energy generation over time (joint power) partially substituted by the hip joint power of the same side. The fact that, even after more than 2 years following the ACL reconstruction and returning to regular activity, the ACL group still had neuromuscular asymmetries suggests a need for improvement in the ACL reconstruction surgery procedures and/or rehabilitation protocols.

On the functional integration between postural and supra-postural tasks on the basis of contextual cues and task constraint

In order to evaluate the effects of uncertainty about direction of mechanical perturbation and supra-postural task constraint on postural control young adults had their upright stance perturbed while holding a tray in a horizontal position Stance was perturbed by moving forward or backward a supporting platform contrasting situations of certainty versus uncertainty of direction of displacement Increased constraint on postural stability was Imposed by a supra-postural task of equilibrating a cylinder on the tray Performance was assessed through EMG of anterior leg muscles angular displacement of the main Joints involved in the postural reactions and displacement of the tray Results showed that both certainty on the direction of perturbation and Increased supra-postural task constraint led to decreased angular displacement of the knee and the hip Furthermore combination of certainty and high supra-postural task constraint produced shorter latency of muscular activation Such postural responses were paralleled by decreased displacement of the tray Thesi results suggest a functional integration between the tasks with central set priming reactive postural responses from contextual cues and Increased stability demand (C) 2010 Elsevier B V All rights reserved

Effects of recovery type after a judo match on blood lactate and performance in specific and non-specific judo tasks

The objective of the present study was to verify if active recovery (AR) applied after a judo match resulted in a better performance when compared to passive recovery (PR) in three tasks varying in specificity to the judo and in measurement of work performed: four upper-body Wingate tests (WT); special judo fitness test (SJFT); another match. For this purpose, three studies were conducted. Sixteen highly trained judo athletes took part in study 1, 9 in study 2, and 12 in study 3. During AR judokas ran (15 min) at the velocity corresponding to 70% of 4 mmol l(-1) blood lactate intensity (similar to 50% (V) over dotO(2) peak), while during PR they stayed seated at the competition area. The results indicated that the minimal recovery time reported in judo competitions (15 min) is long enough for sufficient recovery of WT performance and in a specific high-intensity test (SJFT). However, the odds ratio of winning a match increased ten times when a judoka performed AR and his opponent performed PR, but the cause of this phenomenon cannot be explained by changes in number of actions performed or by changes in match`s time structure.

Coloured Petri nets and graphical simulation for the validation of a robotic cell in aircraft industry

This paper proposes a mixed validation approach based on coloured Petri nets and 3D graphic simulation for the design of supervisory systems in manufacturing cells with multiple robots. The coloured Petri net is used to model the cell behaviour at a high level of abstraction. It models the activities of each cell component and its coordination by a supervisory system. The graphical simulation is used to analyse and validate the cell behaviour in a 3D environment, allowing the detection of collisions and the calculation of process times. The motivation for this work comes from the aeronautic industry. The automation of a fuselage assembly process requires the integration of robots with other cell components such as metrological or vision systems. In this cell, the robot trajectories are defined by the supervisory system and results from the coordination of the cell components. The paper presents the application of the approach for an aircraft assembly cell under integration in Brazil. This case study shows the feasibility of the approach and supports the discussion of its main advantages and limits. (C) 2011 Elsevier Ltd. All rights reserved.

A randomized and controlled trial of a participative ergonomics intervention to reduce injuries associated with manual tasks: physical risk and legislative

A participative ergonomics approach to reducing injuries associated with manual tasks is widely promoted; however only limited evidence from uncontrolled trials has been available to support the efficacy of such an approach. This paper reports on a randomized and controlled trial of PErforM, a participative ergonomics intervention designed to reduce the risks of injury associated with manual tasks. One hundred and seventeen small to medium sized food, construction, and health workplaces were audited by government inspectors using a manual tasks risk assessment tool (ManTRA). Forty-eight volunteer workplaces were then randomly assigned to Experimental and Control groups with the Experimental group receiving the PErforM program. Inspectors audited the workplaces again, 9 months following the intervention. The results showed a significant decrease in estimates of manual task risk and suggested better legal compliance in the Experimental group.

The BCD transient recorder revisited: firmware for the ISA-to-USB bridging card and software for the graphical user interface

This report describes recent updates to the custom-built data-acquisition hardware operated by the Center for Hypersonics. In 2006, an ISA-to-USB bridging card was developed as part of Luke Hillyard's final-year thesis. This card allows the hardware to be connected to any recent personal computers via a (USB or RS232) serial port and it provides a number of simple text-based commands for control of the hardware. A graphical user interface program was also updated to help the experimenter manage the data acquisition functions. Sampled data is stored in text files that have been compressed with the gzip for mat. To simplify the later archiving or transport of the data, all files specific to a shot are stored in a single directory. This includes a text file for the run description, the signal configuration file and the individual sampled-data files, one for each signal that was recorded.

The tasks of medicine: An ideology of care; edited by Peter Baume

No abstract

Contractions of specific abdominal muscles in postural tasks are affected by respiratory maneuvers

The influence of respiratory activity of the abdominal muscles on their reaction time in a postural task was evaluated. The electromyographic (EMG) onsets of the abdominal muscles and deltoid were evaluated in response to shoulder flexion initiated by a visual stimulus occurring at random throughout the respiratory cycle. Increased activity of the abdominal muscles was produced by inspiratory loading, forced expiration below functional residual capacity, and a static glottis-closed expulsive maneuver. During quiet breathing, the latency between activation of the abdominal muscles and deltoid was not influenced by the respiratory cycle. When respiratory activity of the abdominal muscles increased, the EMG onset of transversus abdominis and internal oblique, relative to deltoid, was significantly earlier for movements beginning in expiration, compared with inspiration [by 97-107 ms (P < 0.01) and 64-90 ms (P < 0.01), respectively]. However, the onset of transversus abdominis EMG was delayed by 31-54 ms (P < 0.01) when movement was performed during a static expulsive effort, compared with quiet respiration. Thus changes occur in early anticipatory contraction of transversus abdominis during respiratory tasks but they cannot be explained simply by existing activation of the motoneuron pool.

Deficits on self ordered tasks associated with hyperostosis frontalis interna

A 74 year old patient, EW, with dorsolateral frontal cortical compression due to hyperostosis frontalis interna, in the absence of the Morgagni or Stewart-Morel syndromes, is described. In addition to conventional neuropsychological measures EW was administered one nonspatial and two spatial self ordered working memory tasks, as well as a standard measure of fluid intelligence or g. She showed impaired performance on all three self ordered working memory tasks compared with a normal control group of 10 subjects matched for age, education, sex, and IQ. By contrast, her performance on the fluid intelligence test was comparable with that of the controls. It is concluded that the compression of dorsolateral frontal cortex accompanying hyperostosis frontalis interna may produce selective cognitive impairment.