robot sensing and perception - engineeringpetriu/robot-percep_emil.pdf · school of information...

Emil M. Petriu, Dr. Eng., FIEEEProfessor School of Information Technology and EngineeringUniversity of OttawaOttawa, ON., K1N 6N5 [email protected]

Robot Sensing and Perception

some of the research done in the SMRLab

University of Ottawa School of Information Technology - SITE

Sensing and Modelling Research LaboratorySMRLab - Prof.. Emil M. Petriu



Why natural binary coding cannot be used in practice for absolute position recovery ?A n-bit code would be needed for each quantization step, resulting in n binary tracks in parallel with the guide-path. For instance, the encoding of a 160 m long guide-path with a 0.01 m resolution would need 14 tracks running in parallelwith the guide path

0 0 0 0 1 0 1 0 1 1 1 0 1 1 0 0 0 1 1 1 1 1 0 0 1 1 0 1 0 0 1

0 5 10 15 2520 30p =

PRBS=

A (2n -1) term Pseudo-Random Binary Sequences (PRBS) generated by a n-bit modulo-2 feedback shift register is used as an one-bit / quantization-step absolute code. The absolute position identification is based on the PRBS window property. According to this any n-tuple seen through a n-bit window sliding over PRBS is unique and henceforth it fully identifies each position of the window.

The figure shows, as an example, a 31-bit term PRBS: 0, 0, 0, 0, 1, 0, 1, 0, 1, 1, 1, 0, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 1, 1, 0,1, 0, 0, 1, generated by a 5-bit shift register. The 5-bit n-tuples seen through a window sliding over this PRBs are unique and represent a 1-bit wide absolute position code.



è pseudo-random encoding provides a practicalsolution allowing absolute position recovery withany desired n-bit resolution while employing onlyone binary track, regardless of the value of n.

m•t-1 m •t .... m •t+r ... m • t+t-1 (m+1) • t (m+1) •t+1

Q(m) Q(m+1)

Positions on the scale

Milestones

{x(k)=S(p+n-k)|k=n,...,1} Pseudo-random n-typle corresponding to the position index p = m•t + r

m•t

Control Logic

Shift

Reverse PRBS feedback logic

R(n) R(1)

Stop count

Counter

SEQUENTIAL CODE CONVERSION

. . .

rp = m•t + r

Natural code corresponding to the current position

Adder

. ..

MS

. ..

...

Q(0)

Q(1)

Q(w)

B(n) B(1)

R(n) R(1)

PARALLEL MILESTONE IDENTIFICATION&CONVERSION



Serial-parallel pseudo-random / natural code conversion algorithm



Pseudo-random encoded track (one bit per quantization step) allows recovery of the absolute position of an optically guided Automated Guided Vehicle (AGV)



Optically guided AGV tracking the pseudo-random encoded track



Pseudo-random encoded guide path allows recoveryof the absolute position of an AGV using computer vision



Computer vision recognition of the pseudo-random binary code



Model-based recognition of a pseudo-random encoded object





Wall-mounted pseudo-random encoded guide path allows recovery of the absolute position of the AGV using computer vision



Pseudo-random encoding for computer vision recovery of the 3D position of a probe mapping the electromagnetic–field radiated by a telephone set



Computer vision recovery of the pseudo-random code



0

10

0

0

0

0

0 0

0

1

0

1

0

1

1

1

1

0

0

1

1 1

1 1 1

1

1

1

0 1 1 0 0 1

0 1

1 0

0

0

1

1 0 0 1

0 0

0

1 1

0 1 1 1

0 0

1 10

0 01

j

0 1 2 3 4 5 6 7 8

0

1

2

3

4

5

6

i

Illustrating the window property in a Pseudo-Random Binary Array (PRBA). The 3-by-2 code seen trough a window on a 7-by-9 PRBA is unique and used as absolute code for the window position (i,j).





Pseudo-Random Binary Array (PRBA) encoding for the recovery of the 2D absolute position of a free ranging mobile robot using computer vision



Mobile robot navigation using multiple IR sensors and vision



Path Specifications

ENVIRONMENT

Task

ROBOT

JOINT/WHEEL SERVOS

I.R. RANGE SENSORS

Actuator

Wheel Position

WHEEL/STEER ENCODERS

VISION

ONBOARD COMPUTER

Raster Image

Local Connection

Remote Connection

Object Identities and POSEs

Position Specifications

FRAME TRANSFORMS

TRAJECTORY PLANNER

ROBOT MODELFRAME TRANSFORMS

Robot Position

TASK PLANNER

GEOMETRIC WORLD MODEL

POSITION MODEL

OBJECT RECOGNITION

TRAJECTORY PARAMETRS ESTIMATION

Trajectory Constraints

TELEOPERATOR

VIDEO MONITOR

TAC

TILE

M

ON

ITO

R

&

JOYS

TIC

K

TACTILE SENSOR

COMPOSITE WORLD MODEL

Model-based telepresence control of the mobile robot





Recovery of the 3D shape of objects using structured light



P

P'

P*

PROJECTION GRID

LIGHT SOURCE

CAMERA'S CENTER OF PERSPECTIVE

IMAGE

OBJECT

W

x

y

I

G

x

x

y

y

z

d

CAMERA'S OPTICAL AXIS

PROJECTOR'S OPTICAL AXIS

zz

Q

Point identification problem in pseudo-random encoded structured light



Pseudo-random color encoded structured-lightgrid projected on a 3D object



Recovered corner points at the intersection of grid line edges

Pseudo-random color encodedstructured light grid projected on a cube



Haptic perception of 3D object geometry: the robot arm provides the kinesthetic capability and the tactile sensor probe provides the cutaneous information

Force Sensitive Resistor (FSR) tactile sensor array, 16-by-16 sensing elements on a 1 square inch area



Instrumented passive compliant wrist for tactile exploration of objects



“Tactile display” for computer-human interaction allowing humans to feel through their own touch sense computer-generated geometric profiles. It consists of an array of 8-by-8 vibrators on a 1 square inch area.



robot sensing and perception - engineeringpetriu/robot-percep_emil.pdf · school of information...

Documents