Rakesh Gosangi PRISM lab Department of Computer Science and Engineering Texas A&M University

Whisking with Robots – From Rat Vibrissae to Biomimetic Technology for Active Touch

Tony J. Prescott, Martin J. Pearson, Ben Mitchinson, J. Charles W. Sullivan, and Anthony G. PipeIEEE Robotics & Automation – September 2009.

Rakesh GosangiPRISM lab

Department of Computer Science and Engineering

Texas A&M University

Outline

• Introduction• Rat’s vibrissal system• Neural Processing• Whisking Robots• Discussion and Future

Introduction

• Mammals do a large part of their tactile sensing using vibrissae (whiskers)

• Modern robotics fail to match the capabilities of mammals in tactile perception

Tactile perception in animals

• Alerting stimulus to produce motor response

• Perform complex perceptual tasks like– Determine shape, texture, position of objects in 3-D

space

• Guide motion for nocturnal species like rats and cockroaches

Image – common rat (borrowed from - Whisking with Robots – From Rat Vibrissae to Biomimetic Technology for Active Touch - Tony J. Prescott, Martin J. Pearson, Ben Mitchinson, J. Charles W. Sullivan, and Anthony G. PipeIEEE Robotics & Automation – September 2009.

Tactile perception in animals

• Etruscan shrew prey capture is guided by tactile cues

• Seals can detect hydrodynamic trails left by fish with their whiskers 

Images – water shrew, harbor seal (borrowed from - Whisking with Robots – From Rat Vibrissae to Biomimetic Technology for Active Touch - Tony J. Prescott, Martin J. Pearson, Ben Mitchinson, J. Charles W. Sullivan, and Anthony G. PipeIEEE Robotics & Automation – September 2009.

Tactile Sensors in robotics

• The least trusted sensors– Used as a last line of defense when all the other sensing

modalities fail

• Passive in nature– Waiting to be deflected by an object

Outline

• Introduction• Rat’s vibrissal system• Neural Processing• Whisking Robots• Discussion and Future

Rat’s vibrissal system

• Long facial whiskers called macrovibrissae– Individually actuated and actively controlled by the rats

• Shorter, densely packed microvibrissae on chin and lips– Non actuated

• Mechanoreceptors in the hair follicles convert direction, velocity, duration and torque of whisker movements into electrophysiological signals

Use of macro and micro vibrissae

Image – Use of macro and micro vibrissae (borrowed from - Whisking with Robots – From Rat Vibrissae to Biomimetic Technology for Active Touch - Tony J. Prescott, Martin J. Pearson, Ben Mitchinson, J. Charles W. Sullivan, and Anthony G. PipeIEEE Robotics & Automation – September 2009.

Whisking movements

• Macro vibrissae are moved back and forth (whisking) at high speeds (5-25/sec)

• The movement of the whiskers is controlled depending on head-body movement, recent sensory experience

Movement of whiskers of a head-restrained rat

Image – Whisking control in rat (borrowed from - Whisking with Robots – From Rat Vibrissae to Biomimetic Technology for Active Touch - Tony J. Prescott, Martin J. Pearson, Ben Mitchinson, J. Charles W. Sullivan, and Anthony G. PipeIEEE Robotics & Automation – September 2009.

Whisking movements

• Whiskers move asymmetrically when the rat turns its head

• Rats can control the speed of the individualwhiskers

• Whiskers have many degrees of freedom– Parallel and perpendicular to the plane of the head– Torsional rotation

Image – Asymmetry in whisking movements (borrowed from - Whisking with Robots – From Rat Vibrissae to Biomimetic Technology for Active Touch - Tony J. Prescott, Martin J. Pearson, Ben Mitchinson, J. Charles W. Sullivan, and Anthony G. PipeIEEE Robotics & Automation – September 2009.

Outline

• Introduction• Rat’s vibrissal system• Neural Processing• Whisking Robots• Discussion and Future

Neural Processing

• Whisker deflections are converted into physiological signals.

• These signals are processed in the thalamus and sensory cortex

• There exists a one-to-one mapping between the whiskers and barrels inthe sensory cortex

Image – Vibrissal sensory processing pathway (borrowed from - Whisking with Robots – From Rat Vibrissae to Biomimetic Technology for Active Touch - Tony J. Prescott, Martin J. Pearson, Ben Mitchinson, J. Charles W. Sullivan, and Anthony G. PipeIEEE Robotics & Automation – September 2009.

Outline

• Introduction• Rat’s vibrissal system• Neural Processing• Whisking Robots• Discussion and Future

Whisking Robots

• Sensor transduction– Potentiometers – to measure torque– Electret microphones

• Sensitive to deflection but cannot detect direction– Piezoelectric sensors

• Cannot measure static deflections– Magnetic Hall-effect sensors

• Robust, lightweight and sensitive

Whisking Robots

• Actuation– Independently actuated whiskers– Uniform actuation of whiskers

• Mechanical properties of the vibrissal shaft– Steel wires– Molded composites

• Morphed like rat whiskers

• Signal processing– Neuromorphic algorithms

aMouse

• Rat whiskers were glued to electret microphones• ANNs and spectral analysis for signal processing

Image – aMouse (borrowed from - Whisking with Robots – From Rat Vibrissae to Biomimetic Technology for Active Touch - Tony J. Prescott, Martin J. Pearson, Ben Mitchinson, J. Charles W. Sullivan, and Anthony G. PipeIEEE Robotics & Automation – September 2009.

Whisking sensobot

• 4x1 array of whiskers with strain gauges– Extract radial distance– Estimate 3-D object shape

Image – Whisking sensobot (borrowed from - Whisking with Robots – From Rat Vibrissae to Biomimetic Technology for Active Touch - Tony J. Prescott, Martin J. Pearson, Ben Mitchinson, J. Charles W. Sullivan, and Anthony G. PipeIEEE Robotics & Automation – September 2009.

Darwin IX

• Whiskers detect deformation along their length• Employed neuromorphic computational methods

– Texture discrimination

Image – Darwin IX (borrowed from - Whisking with Robots – From Rat Vibrissae to Biomimetic Technology for Active Touch - Tony J. Prescott, Martin J. Pearson, Ben Mitchinson, J. Charles W. Sullivan, and Anthony G. PipeIEEE Robotics & Automation – September 2009.

Whiskerbot

• Orienting to the detectedtargets

Image – Whikserbot (borrowed from - Whisking with Robots – From Rat Vibrissae to Biomimetic Technology for Active Touch - Tony J. Prescott, Martin J. Pearson, Ben Mitchinson, J. Charles W. Sullivan, and Anthony G. PipeIEEE Robotics & Automation – September 2009.

Scratchbot

• Increased degrees of freedom for moving and positioning the whiskers

• Including a neck with three degrees of freedom• Hall-effect sensors

Image – Scratchbot (borrowed from - Whisking with Robots – From Rat Vibrissae to Biomimetic Technology for Active Touch - Tony J. Prescott, Martin J. Pearson, Ben Mitchinson, J. Charles W. Sullivan, and Anthony G. PipeIEEE Robotics & Automation – September 2009.

BIOTACT

• Modular vibrissal sensing units to be assembled into different configurations

Image – Biotact (borrowed from - Whisking with Robots – From Rat Vibrissae to Biomimetic Technology for Active Touch - Tony J. Prescott, Martin J. Pearson, Ben Mitchinson, J. Charles W. Sullivan, and Anthony G. PipeIEEE Robotics & Automation – September 2009.

Outline

• Introduction• Rat’s vibrissal system• Neural Processing• Whisking Robots• Discussion and Future

Discussion

• Tactile sensing based navigation is useful in visually occluded environments

• Also useful in texture and shape recognition

• Construction of 3-D tactile maps of the environment

• The transducers (receptors) are away from contact surface– No damage due to direct physical contact

Challenges

• Better understanding of sensory motor loops in the vibrissal systems of mammals

• Biomimetic algorithms for whisker control and processing vibrissal signals

• Mapping whisker deflection signals to surface and shape properties

Questions / Comments