Representation and Experience-Based Learning of Explainable Models for Robot Action Execution

Author(s):  
Alex Mitrevski ◽  
Paul G. Ploger ◽  
Gerhard Lakemeyer
2018 ◽  
Vol 1 (2) ◽  
pp. 57
Author(s):  
Masadah Masadah

Education is a conscious and systematic effort not only to humanize human beings but also for human beings to realize their position as khalifatullah fil ardhi, which in turn will increasingly increase itself to be a pious, faithful, knowledgeable and virtuous man. In general the problems formulated in this research is whether Implementation of Contextual Learning with Learning Community approach can improve student's motivation and achievement in FIQH study field? How Implementation of Contextual Learning has a Learning Community approach that can improve students' motivation and achievement in FIQH? Field. This research was conducted in Mojokerto Regency, precisely at MI Mambaul Hidayah Mengelo Sooko Mojokerto. This research is a classroom action research with collaborative type. This research phase follows a model developed by Kemmis and Taggart, which is a spiral cycle that includes planning activities, action execution, observation, and reflection. The data collection techniques used are: (1) observation; (2) measurement of learning result test; and (3) documentation. Data obtained from the action are then analyzed. Qualitative data consisting of observation and documentation are analyzed qualitatively, while data collected in the form of numbers or quantitative data, simply by using descriptive analysis and visual presentation. Based on the results of research that has been implemented can be concluded that the Implementation of Contextual Learning with Learning Community approach can improve student's motivation and achievement in the field of FIQH study. From the data in the field shows that there is an increase in student learning motivation that the initial average value of pre-test of 20 increased to 24 or about 20% in cycle I, in cycle II more increased to 31 or about 55%, and in cycle III the more increased to 45 or about 125%. Level of increase between cycle I with cycle II about 29%, between cycle II with cycle III about 45%, between cycle III with cycle I about 87%. With the increase of students' learning motivation, their learning achievement also increased, whereas the average value of pre test of 6.60 increased to 6.84 or about 4% in cycle I, in cycle II more increased again to 7.75 or about 17 %, and in cycle III it increases to 8.80 or about 35%. The level of improvement between cycle I with cycle II is about 13%, between cycle II with cycle III about 15%, between cycle III with cycle I about 30%.


2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Stefano Rozzi ◽  
Marco Bimbi ◽  
Alfonso Gravante ◽  
Luciano Simone ◽  
Leonardo Fogassi

AbstractThe ventral part of lateral prefrontal cortex (VLPF) of the monkey receives strong visual input, mainly from inferotemporal cortex. It has been shown that VLPF neurons can show visual responses during paradigms requiring to associate arbitrary visual cues to behavioral reactions. Further studies showed that there are also VLPF neurons responding to the presentation of specific visual stimuli, such as objects and faces. However, it is largely unknown whether VLPF neurons respond and differentiate between stimuli belonging to different categories, also in absence of a specific requirement to actively categorize or to exploit these stimuli for choosing a given behavior. The first aim of the present study is to evaluate and map the responses of neurons of a large sector of VLPF to a wide set of visual stimuli when monkeys simply observe them. Recent studies showed that visual responses to objects are also present in VLPF neurons coding action execution, when they are the target of the action. Thus, the second aim of the present study is to compare the visual responses of VLPF neurons when the same objects are simply observed or when they become the target of a grasping action. Our results indicate that: (1) part of VLPF visually responsive neurons respond specifically to one stimulus or to a small set of stimuli, but there is no indication of a “passive” categorical coding; (2) VLPF neuronal visual responses to objects are often modulated by the task conditions in which the object is observed, with the strongest response when the object is target of an action. These data indicate that VLPF performs an early passive description of several types of visual stimuli, that can then be used for organizing and planning behavior. This could explain the modulation of visual response both in associative learning and in natural behavior.


2012 ◽  
Vol 231 (1) ◽  
pp. 124-129 ◽  
Author(s):  
Michela Loporto ◽  
Craig J. McAllister ◽  
Martin G. Edwards ◽  
David J. Wright ◽  
Paul S. Holmes

Author(s):  
Vittorio Gallese ◽  
Corrado Sinigaglia

Mental simulation was claimed to provide a distinctive way of gaining knowledge about others’ actions and thoughts since the late 1980s. A decade later, the discovery of mirror neurons in macaque monkeys and the evidence of mirror brain areas in humans presented a new angle on this claim, suggesting also an embodied approach to simulation. The aim of the present chapter is to introduce and discuss this embodied approach and its role in basic social cognition. In doing this, we shall start by characterizing the distinctive features of embodied simulation (ES), especially in relation to its its motor aspects. Then, we shall provide evidence for the claim that ES may be critically involved in understanding others’ actions. Finally, we shall explore the conjecture that ES might involve a common ground for action execution and observation not only at the functional but also at the phenomenological level.


Author(s):  
Stephan de la Rosa ◽  
Frieder L. Schillinger ◽  
Heinrich H. Bülthoff ◽  
Johannes Schultz ◽  
Kamil Uludag

2018 ◽  
Vol 8 (8) ◽  
pp. 1261 ◽  
Author(s):  
René Schmidt ◽  
Alexander Graf ◽  
Ricardo Decker ◽  
Verena Kräusel ◽  
Wolfram Hardt ◽  
...  

Achieving lightweight construction through only material substitution does not realize the full potential of producing a lightweight material, hence, it is no longer sufficient. Weight-saving goals are best achieved through additional function integration. In order to implement this premise for mass production, a manufacturing process for joining and forming hybrid laminates using a new tool concept is presented. All materials used are widely producible and processable. The manufactured cover of an automotive center console serves to demonstrate a human interface device with impact detection and action execution. This is only possible through a machine learning system, which is implemented on a small—and thus space- and energy-saving—embedded system. The measurement results confirm the objective and show that localization was sufficiently accurate.


Sign in / Sign up

Export Citation Format

Share Document