understanding mirror neurons : a bio-robotic approach
DESCRIPTION
Understanding mirror neurons : A bio-robotic approach. 인지과학 협동과정 VCC 석사 2 학기 김시온. 1.Introduction. Understanding action. 사 과. 때리다. 왜 ?. 미워서 , 웃겨서 , 짜증나서 까불어서 , 사랑의 마음으로 잘되라고 , 말 안들어서 …. -> 의도 파악의 필요성. 1.Introduction. Speech understanding machine. - PowerPoint PPT PresentationTRANSCRIPT
Understanding mirror neurons: A bio-robotic approach
인지과학 협동과정VCC 석사 2 학기 김시온
Understanding action1.Introduction
사과 때리다왜 ?
미워서 , 웃겨서 , 짜증나서까불어서 , 사랑의 마음으로잘되라고 , 말 안들어서…-> 의도 파악의 필요성
Speech understanding machine
• Alvin Liberman(May 10, 1917 - Jan. 13, 2000)
1.Introduction
Analyzing the acoustic characteristics of spoken words
Failed because of the limited computational power available
Subjects Context
Sound Articulatory gestures
Acoustically –evoked motor “Resonance”1.Introduction
Perception 에도 관여한다 .
Production 에 관여한 지식들이
Action 과 관련하여 이러한 역할을 하는 것으로보이는 영역이 발견되었다 !!
2. Physiological properties of monkey rostroventral premotor area (F5)
Inferior premotor area
2. Physiological properties of monkey rostroventral premotor area (F5)
Motor Neurons
• Goal-Directed action 들에 반응 (grasping, manipulating, tearing, holding) - 비슷한 손동작이 사용될지라도 목적이 다르면 발화하지 않음• 많은 Grasping neuron 들은 특정한 종류 의 Grasping 에만 선택적 반응 (precision, finger prehension, whole hand grasping)
-> a repository of motor actions
Visuomotor Neurons- Canonical Neurons
• 움직임이 있기 전에 , 움직일 수 있는 대상을 보았을 때에도 발화• 많은 canonical neurons 들은 하나 , 혹은 소수의 대상에 선택적으로 반응한다 .
2. Physiological properties of monkey rostroventral premotor area (F5)
Visuomotor Neurons- Mirror Neurons
• 내가 행동할 때 뿐만 아니라 , 다른 행위자가 동일한 목표 - 지향적 행동을 할 때에도 발화• 최근 연구 결과 F5 이외에도 PF 에서도 발견됨• 다른 행위자가 손이나 입을 이용해 물체와 상호작용할 때 활발하게 반응• 도구를 이용했을 때에는 반응하지 않음• 대부분의 Mirror Neuron 들은 한 종류의 행동에만 선택적으로 반응 (Highly congruent neurons)
2. Physiological properties of monkey rostroventral premotor area (F5)
Visuomotor Neurons- Mirror Neurons
2. Physiological properties of monkey rostroventral premotor area (F5)
A role of F53. A model of area F5 and the mirror system
ControllerClassifier
Object affordance
Visual appearance of the hand
Controller-predictor formulation3. A model of area F5 and the mirror system
Acting Specific Motor Plan
Desired action, timing
Context
※ Feedback : to counteract disturbances and to learn from mistakes
ComparisonF5 forward information
≒ Sensory feedback(VMM)
Desired action
Controller-predictor formulation3. A model of area F5 and the mirror system
Observing an action Transforming visual cues into motor information as before
Mirror Neurons
Bayesian approach??
Ontogenesis of mirror neurons
• 아기들이 행위자의 목표를 판단할 때 사용하는 Cue 는 ? (Path? Object?)
• Results
6,9 mos : new>old goal
Woodward(1998)
3. A model of area F5 and the mirror system
3. A model of area F5 and the mirror system
1) Certain actions are more likely to be applied to a particular object2) Objects are used to perform actions
How do we achieve an action recognizing?
Canonical Neuron
Mirror N
euronLinkingParticular action= particular effect
Specific goals
Question!!
• 과연 VMM 정보가 Action 인식에 도움을 줄까 ??
• 로봇를 가지고 한번 직접 실험해 보자 !
4. A machine with hands
Recording “motor awareness”
• Cyber-glove, a pair of CCD cameras, a magnetic tracker, two touch sensors
• Data was sampled at frame rate, synchronized and stored to disk by a Pentium class PC
• Cyber-glove has 22 sensors , recording the kinematics of the hand at up to 112Hz
• Two touch sensors were mounted on the thumb and index finger to detect the moment of contact with the object
• Matlab was employed for post-processing
Motor-kinesthetic information + visual information vs only visual information
Grasping type followed Napier’s taxonomy4. A machine with hands
4. A machine with hands
• Each grasping action was recorded from six different subjects.• Moving the cameras to 12 different locations• Images of the scene from the two camera synchronized with the cyber-glove and magnetic tracker data
• The visual features were extracted from pre-processed image data• Using a simple color segmentation algorithm• The data set was filtered through Principal Component Analysis
=> Bayesian Classifier
Classification Task
Motor-kinesthetic information + visual information vs only visual information
VMM
• Using a simple back-propagation neural network with sigmoidal units
• The input of the VMM was the vector of the mapping of the images onto the space spanned by the first N PCA vectors
• The output was the vector of joint angle acquried the data glove.
4. A machine with hands
Experiment 1 - Classification4. A machine with hands
>-> 운동정보가 Action recognition 에 도움을 줄 수 있다 !!!
Cog5. Robotic experiment
Flipper
A mirror neuron-like representation could be acquired by simply relying on the information exchanged during the robot- environment interaction( 로봇 - 환경 간의 상호작용에서 mirror-neuron 과 같은 표상을 획득할 수 있는가 ?)
Causal indirection, brain area and function5. Robotic experiment
If any neural unit is active in these two situations(both when acting and observing)Then it can be regarded in all respects as a “mirror” unit.
1
by repetitive exposure and manipulation
Rolling affordances6. Learning object affordances
Grasping(x)Touching, poking, prodding, sweeping(o)
6. Learning object affordances
Rolling affordances
Generic poking actions
Can this information be re-used when acting to generate anything usefulShowing exploitation of the object affordance?
6. Learning object affordances
100/15 … 3 번만 완전 실패 ( Canonical Neuron)≒Another interesting question is whether knowledge about object directed actionscan be reused in interpreting observed actions performed by a human experimenter.
mimicry7. Developing mirror neurons
Same segmentation procedure section 6 generated by human and robot
… LEARNING??
Passive vs Active imitation
Demiris & Hayes, 2002
7. Developing mirror neurons
mimicry7. Developing mirror neurons
Learning of mirror neurons can be carried out by a process of autonomous development
• The motor information plays a role in the recognition process
• A mirror-like representation can be developed autonomously on the basis of the interaction between an individual and the environment
7. Conclusion
감사합니다 .