ROBOT CONTROL USING LIVING ROBOT CONTROL USING LIVING NEURONAL CELLS: PROGRESS AND CHALLENGES Vi Victor M. Becerra M B School of Systems Engineering Whitehead Lecture Series G ld Goldsmiths, 09 March 2011 i h M h
Introduction Introduction
Background Background � The biological brain can be seen as a complex Th bi l i l b i b l computational platform � Progress is being made towards hybrid b d d h b d systems that integrate biological neurons and electronic components. l t i t
Detached brains Detached brains � A group from NW University g p y interfaced the detached brain of a lampray with a mobile robot (Reger et al 2000) robot (Reger et al , 2000) � Robot light sensors provide stimulation to the brain. stimulation to the brain. � Wheel motors activated by signals derived from neural activity. � Showed stable behaviours, and adaptation to changes in and adaptation to changes in sensory input
Living creatures Living creatures � A group from NY State University remotely controlled the motion of a rat (Talwar et al , ll d h f l l 2002) � Direct stimulation of the rat’s brain using electrodes l t d � Rat trained to interpret stimuli as cues stimuli as cues. � Rewards given as separate electrical separate electrical stimuli. Source: http://www.nature.com/nature/journal/v417/n6884/full/417037a.html
Issues Issues � Detached whole brains are difficult to keep alive alive. � Interfaces to the brain in living beings are problematic due to barriers (e.g. skin, skull) problematic due to barriers (e.g. skin, skull) � They may be invasive and destructive, leading to potential ethical problems. g p p � Data interpretation is difficult given the large number of neurons. � Data collection usually restricted to few regions of the brain.
Cultured neurons Cultured neurons � Neurons cultured in laboratory conditions and interfaced using a planar multi electrode array interfaced using a planar multi ‐ electrode array (MEA) � Non ‐ invasive approach o as e app oac � Allows probing the operation of biological neuronal networks � Allows measuring activity in the whole structure
The cultures The cultures � Created by dissociating neurons taken from cortical t k f ti l tissue from foetal rodents using enzymes. g y � Cultured in a special chamber and provided with suitable environmental bl l conditions and nutrients. � Spontaneous reconnection � Spontaneous reconnection to nearby neurons in a short period.
Importance of this type of Importance of this type of study � Better understanding of interaction between the brain and external devices the brain and external devices. � Understanding the relations between neuronal activity and behaviour is critical for neuronal activity and behaviour is critical for dealing with neurological disorders. � Re embodiments (real or virtual) may help � Re ‐ embodiments (real or virtual) may help the study of biological learning mechanisms.
Related work – simulated Related work simulated robot � D M � DeMarse et al (2001) and Shkolnik (2003) t l ( ) d Shk l ik ( ) interfaced a neuronal culture with a simulated mobile robot simulated mobile robot � A MEA was employed in both cases � Patterns of the electrical activity of the Patterns of the electrical activity of the network were interpreted as robot commands. � DeMarse and co ‐ workers provided electrical stimulation based on information from the simulated environment i l d i
Related work – aircraft Related work aircraft simulator � DeMarse and Dockendorf � DeMarse and Dockendorf (2005) interfaced a neuronal culture with a simulated aircraft simulated aircraft � A MEA was employed � The weights of the network were manipulated through were manipulated through stimulation. � The living network was used to control pitch and d t t l it h d roll of the simulated Source: http://neural.bme.ufl.edu/page12/page1/page1.html aircraft.
The work at Reading The work at Reading � EPSRC funding 2007 ‐ 2010 EPSRC f di � Collaboration between Systems Engineering and Pharmacy d h � Four academics � One RA and three PhD students
The work at Reading The work at Reading � Hypothesis: Disembodied biological networks H h i Di b di d bi l i l k must develop within a closed loop sensory interaction with the environment The loop interaction with the environment. The loop may be closed with a robot. � Supported by studies that show that S t d b t di th t h th t development in a sensory deprived environment results in dysfunctional neural environment results in dysfunctional neural circuitry
Experimental setup ‐ MEA Experimental setup MEA � The MEA enables voltage to be recorded at 59 out of 64 electrodes. � Detection area 100 μ m around each electrode, which have a radius of 30 μ m � Sampling frequency of 25 kHz p g q y 5 � Software allows to separate the firing of small groups of neurons from an electrode � The same electrodes can do stimulation e sa e e ect odes ca do st u at o � A picture of the global activity of the network can be formed � Culture population order around 10 4 ‐ 10 5 Culture population order around 10 10 neurons
The measured neural activity The measured neural activity � Linux based open source MEAbench software
Stimulation Stimulation � Stimulation software developed at Reading � Stimulation software developed at Reading
Experimental setup ‐ robot Experimental setup robot � Two motor ‐ driven T d i wheels � Sonar sensors to S detect obstacles � Wireless communications � Programmable on ‐ board processor
Experimental setup: Experimental setup: processing � Signal processing can be broken into two sections: sections: � Culture ‐ to ‐ robot: neuronal activity is procesed and features of interest are mapped into robot pp commands � Robot ‐ to ‐ culture: the robot sensor readings are mapped into stimulus to the culture
Experimental setup: Experimental setup: Hardware � Head stage connecting to the MEA H d i h MEA � 60 channel amplifier � PC data acquisition card � Stimulus generator � Workstation acquires and processes neural data � PC runs robot control software � Miabot robot platform Miabot robot platform
Experimental setup Experimental setup
Experimental setup ‐ detail Experimental setup detail
Experiments: Experiments: Input ‐ output pair selection � Suitable neuronal pathways were sought. S i bl l h h � Biphasic pulses were applied with a magnitude of 600 mV 100 μ s each phase repeated 16 times 600 mV, 100 μ s each phase, repeated 16 times. � Input ‐ output pairs were defined as follows: � Activity measured in electrode “B” in response to � Activity measured in electrode B in response to stimulus in electrode “A”, within 100 ms � Electrode “B” responds more than 60% of the Electrode B responds more than 60% of the time to electrode “A”, and less than 20% to stimulation in other electrodes.
Experiments: standard control Experiments: standard control
Experiments: culture in the loop Experiments: culture in the loop
Experiments Experiments Sonar measured distance to the wall � When the robot approaches a wall, the distance (green line) decreases below a threshold (30 cm), triggering a stimulation pulse pulse. � Significant activity events at the output electrode are translated into a 90 degree turn (rotation starts on the yellow line and stops on the red line) line, and stops on the red line). � Some random turns were registered due to the network spontaneous activity
Experiments: statistics Experiments: statistics � 67% stimulation – response event 6 % i l i � Run time: 140 s � Total closed loop time: 0.2 ‐ 0.5 s � Meaningful turns: 46% (s.d. 15%) � Spontaneous turns: 54% (s.d. 19%)
Studying the culture Studying the culture � In order to better understand the activity of I d b d d h i i f the neural culture and its evolution over time (which should help to improve the closed loop (which should help to improve the closed loop scheme), we have done the following work: � Modelling the network states transition patterns � Modelling the network states transition patterns � Investigating the functional connectivity of the network and its evolution over time network and its evolution over time
Modelling the network state Modelling the network state transition patterns � We have used Hidden Markov Models, which W h d Hidd M k M d l hi h are probabilistic models of state transitions. � Models trained on MEA data with stimulus d l d d h l � Network discrete ‘observables’ were defined on the basis of the presence of a spike on a given channel during a short time interval (0.1 ms). )
Modelling the network state Modelling the network state transition patterns
Modelling the network state Modelling the network state transition patterns State flow diagram of state transitions
Recommend
More recommend
