MOSS - Multi Ocular Smart System
Via Provinciale Pianura 2, Zona Industriale San Martino int 23 – 80078 Pozzuoli (Naples) ITALY
Tel. +39 0815263475 Fax +39 0815262701 - e.mail : info@tsd-space.it - web page : www.tsd-space.it
MOSS
MOSS FEATURE HIGHLIGHTS – Historical Background
The MOSS project, carried out by Techno System Developments and co-financed by the Italian
Space Agency, is aimed to develop a compact and high performances equipment for visionbased navigation.
Vision-based navigation can be considered as a major enabling technology in support of the
autonomy requirements of space applications like: Exploration missions (landers, rovers, etc.),
On-Orbit Servicing Applications (like Satellite Inspections, Rendezvous, Docking etc.) and
Formation-Flying Missions.
Future space missions certainly will ask for an increased degree of autonomy on board, and
consequently, the requirements for visual navigation shall become increasingly demanding
The MER (Mars Exploration Rover) mission that landed two rovers (Spirit and Opportunity) on
Mars, in 2004, represents the first mission using vision-based navigation techniques for
landers and for rovers like stereovision, visual odometry, and feature detection and tracking
Multi Ocular Smart System
2
MOSS
MOSS FEATURE HIGHLIGHTS – Historical Background
MER was a very successful mission and the validity of the visual navigation was fully
demonstrated, but:
Limited performances
of the adopted
imaging/processing
equipment
very low rover
speed
Low amount of distance and scientific
data in a given time frame
It was clear that for the subsequent mission it would be necessary to increase the rover
speed; so higher performances electronics for faster implementations of vision and
planning algorithms were considered a very important and priority issue
New and more powerful technologies have been in fact adopted by NASA for the Mars
Science Laboratory (MSL) mission, currently on going, that landed on Mars, on August 6,
2012, the rover “Curiosity” (RAD 750 CPU providing 400MIPS instead of RAD6000 of MER
providing only 35MIPS, 2Mpixel camera instead of 1Mipixel, motorized lens for zooming
and autofocus, real time compression, etc.)
Multi Ocular Smart System
3
MOSS
MOSS FEATURE HIGHLIGHTS – Historical Background
Traditionally:
image processing architectures
centralized and based on a
single microprocessor
navigation algorithms
implemented at software
Low performances of
microprocessors
compatible with space
environment
very low execution rate
not real-time
performances
In most of the cases
image processing and
rover locomotion not
performed at the
same time
For future missions the imaging/processing equipment
shall be able:
to offer faster image processing, compatible with
the computational loads of more sophisticated
navigation algorithms
to fulfill the constraints in terms of mass, volume,
and power consumption of the missions increasingly
based on small platforms
Multi Ocular Smart System
4
MOSS
MOSS FEATURE HIGHLIGHTS – General System Architecture
The MOSS is a high-modular and flexible system that can be used in different configurations by
combining the following parts:
Multi-ocular Camera (Multi-Cam) provided with three fixed lens
CMOS Monocular Camera (CMOS Mono-Cam) provided with one fixed lens
CCD Monocular Camera integrated with motorized lens
High-performance Processing Unit for Visual Navigation (HPVN)
The HPVN is dedicated to Image processing and comprises the following modules:
Image Processing Module (IPM) composed of:
o a mother board for the implementation of the unit control & communication
functionalities, and the high-level image processing algorithms for visual navigation
o a mezzanine board for the intensive-computing required by the algorithms
Power Conditioning & Distribution Module (PCDM)
The cameras were designed to be operated also in a stand-alone configuration, without requiring the
use of the HPVN. The cameras are also internally provided with a lossy compressor of the images,
based on Wavelet SPHIT algorithm, and, in stand-alone configuration, they can output compressed
video by means of an embedded SpaceWire I/F.
Multi Ocular Smart System
5
MOSS
MOSS FEATURE HIGHLIGHTS – Operational Scenarios & Applications
Camera System
Camera
Multi-Cam
Volume: 230 x 78 x 49.5 mm3
(excluding optics)
Mass: 880 g (excluding optics)
1250 g (including optics)
CMOS Mono-Cam
Volume: 69.5 x 69.5 x 55 mm3
(excluding optics and fixation
points)
Mass: 310g (excluding optics)
465 g (including optics)
CCD Mono-Cam
Volume: 219.75 x 127.9 x 105
mm3 (excluding mounting feet)
Mass: 2500 g
Optical Parameters
3D Camera
FOV = ±45° (3D Vision)
WD = From 0.6 m to 5 m
Rover navigation
On-Orbit servicing (close range manoeuvring and
docking)
Panoramic Camera
WD = From 4m to infinity
FOV = ±15° (Panoramic)
Rover panoramic view
On-Orbit servicing (medium range manoeuvring)
FOV = ±50°
WD = From 1m to infinity
FOV = ±25° @17mm to
±7°@65mm
WD = From 5m to infinity
Processing Unit for Vision
based Navigation
Specifications
HPVN
Volume:249.3 x 177 x 55.2 mm3
(excluding mounting feet)
Mass: 2255 g
Power: 28V ± 25%
Comm. & Control:6.9W@28V
Processing Section:
1FPGA: 15.1W @28V
2 FPGAs: 23.3W@28V
Real-Time Stereo vision
Wavelet based Image compression
@30fps, CF=(8 ÷100) in lossy o
loss less mode
Feature extraction & tracking
Open Architecture for custom
algorithms development
Multi Ocular Smart System
Applications
6
Landing
On-Orbit servicing (medium range
manoeuvring)
On-Orbit servicing (medium-far range for
localization manoeuvring)
Applications
Landing
Rovering
Rendez-vous manoeuvring
MOSS
MOSS FEATURE HIGHLIGHTS – Possible System Configurations and
Qualification levels
The System is highly modular and
components can be combined so
implement different configurations,
accordance with the requirements
different operational scenarios
its
to
in
of
The fault tolerance grade embedded in
some part of the system and the
extensive modularity give the possibility
to differently combine the parts, so to
fulfill specific redundancy requirements
Significant efforts have been done in the
selection of the components and
identification of proper design solutions,
so to offer the system in two different
space qualification levels:
Fully ITAR free, and low cost, with
a radiation tolerance of at least
10Krad
ITAR version with a radiation
tolerance of at least 100Krad
Multi Ocular Smart System
7
MOSS
Technological Aspects related to component selection and qualification
As previously mentioned FM design of the MOSS is available in two different versions:
full ITAR-free and with radiation tolerance of at least 20 Krad
subject to ITAR and with radiation tolerance of at least 100 Krad.
To fullfil the full ITAR-free version requirement, TSD has carried out 6 radiation test campaigns
involving a remarkable number of components and in particular 30 components of different
typology (main listed below):
Image Sensors (CCD and CMOS active pixel sensors)
Memories (PROM, Flash Memory, SDRAM)
Transceivers and SerDes
Power Devices (DC/DC Converters, Switching Regulators, Dual Full-Bridge PWM Motor
Driver, Mosfets)
A/D converters and D/A converters
Each test was carried out with an irradiation up to a total dose of at least 30 kRad and required
the development of a specific HW and SW
Multi Ocular Smart System
8
MOSS
Technological Aspects related to component selection and qualification
The HPVN modules have been designed so to perform computational intensive tasks while
keeping mass and power consumption low.
To comply with this requirement, the usage of FPGA is the preferred solution. Indeed, the
availability of high-density devices allows the designer to integrate most of the functions in only
one component, reducing component count, decreasing board power consumption, avoiding
bottleneck in off-chip data transfer, supporting hardware acceleration of computationally
intensive tasks.
Antifuse, flash and reprogrammable FPGAs are today available on the market for space
application, with different specifications in terms of reliability and performances. HPVN
employs all these technologies to best suit with any specific requirements of each single
module.
Hardware acceleration of video processing functions is implemented in the recently released
space-grade Xilinx Virtex-5 QV devices while CPU IP cores for control and configuration
functions are fitted into Actel radiation tolerant Axcelerator FPGAs.
Camera modules are based on flash devices, providing the best compromise between
performances, power consumption, availability (live at power-up) and reliability.
Multi Ocular Smart System
9
MOSS
Multi--Ocular Camera
Multi
Image size &
Frame rate
Multi--cam 3D model
Multi
1280x720 (720p) @ 30fps for the panoramic image
1024x1024 @10fps for the 3D vision
Redundant power Input Interface
Uncompressed Video data I/F: RS-644 LVDS SerDes
Compressed Video data interfaces: RS-644 LVDS,
SpaceWire
I/O
Configuration & Control I/F: Redundant CAN bus
External trigger input I/F (Left & Right): RS-644
(LVDS)
Thermal control Interfaces: 2 Heaters + 5
Thermistors
Power supply: 5V ± 5 %
Stereovision only: 3.3W@5V
Power
Panoramic image acquisition only: 3.22W@5V
Characteristics
Full functions activated: 7.3W@5V
Panoramic Image Compression: 5.2W@5V
Volume
230 x 78 x 49.5 mm3 (excluding optics)
880 g (excluding optics)
Mass
1250 g (including optics)
±45° (3D Vision)
FOV
±15° (Panoramic)
Multi-Cam Main
MultiSpecifications
Multi Ocular Smart System
10
MOSS
Multi--Ocular Camera Design Solutions
Multi
Three CMOS image sensors:
2 B/W, 1Mpixels@10fps, CMOS
APS for 3D vision
1 HD1080p
(1920x1080pixel@30fps) color
CMOS APS for panoramic vision
Architecture based on three
sections:
Main Electronics
Focal Plane proximity
electronics
Power Electronics
All three sections assembled on one
rigid-flexible PCB thus offering:
Thermal and mechanical
decoupling of different sections
Absence of internal connectors
Independent heater and temperature sensor for each image sensor
Fault tolerance for the main functionalities (FPGA, power sections, communication I/Fs, etc.)
Multi Ocular Smart System
11
MOSS
CMOS Monocular Camera
Image size & 1280x720 (720p) @ 60fps
Frame rate
1920x1080(1080p) @30fps
Block diagram
Power Input Interface
Uncompressed Video Data I/F: RS-644 LVDS
SerDes
Compressed Video Data I/F: RS-644 LVDS
SpaceWire
I/O
Configuration & Control I/F: Redundant CAN
bus Interfaces
External trigger input I/F: RS-644 LVDS
Thermal control Interfaces: 1 Heater + 2
Thermistors
Power supply voltage:
5V ± 5 %
Acquisition only (1080p):
2.3W@5V
Power
Acquisition & Compression(720p):
Characteristics 3.5W@5V
Acquisition only (720p):
2.2W@5V
69.5 x 69.5 x 55 mm3 (excluding optics and
Volume
fixation points)
310g (excluding optics)
Mass
465 g (including optics)
FOV
±50°
CMOS MonoMono-Cam 3D model
Multi Ocular Smart System
12
CMOS MonoMono-Cam Main
Specifications
MOSS
CCD Monocular Camera
Motorized optic offering zoom, autoiris, and autofocus
Opto-mechanical and electronics parts integrated in the same volume to minimize mass and volume
Image size &
1920x1080 (1080p) @30fps
Frame rate
CCD MonoMono-Cam
3D model
Mass
FOV
Power Input Interface
Uncompressed video data I/F: RS-644 LVDS
SerDes
Compressed video data I/F: RS-644 LVDS
SpaceWire
Configuration & Control: Redundant CAN
bus Interfaces
External trigger input I/F: RS-644 (LVDS)
Thermal control Interfaces: 2 Heaters + 2
Thermistors
Power supply voltage: 28 V ± 25 %, 5V ± 5 %
Acquisition only: 1.8W@ 5V; 4.1W@28V
Acquisition & Compression: 1.9W@ 5V;
4.1W@28V
219.75 x 127.9 x 105 mm3 (excluding
attachment points)
2500 g
from ±25°@17mm to ±7°@65mm
Motorized
Lens
2 stepper motors for focus and zoom and 1
DC motor for iris
I/O
Power
Characteristics
Volume
Block diagram
Multi Ocular Smart System
CCD MonoMono-Cam Main Specifications
13
MOSS
HPVN
The HPVN is dedicated to the image processing
and communication with the spacecraft C&DH;
it comprises the following modules:
Image Processing Module (IPM)
Power Conditioning & Distribution
Module (PCDM)
The IPM is composed of four sections:
Image Processing A
Image Processing B
Data Handling A
Data Handling B
Two Image Processing sections and two Data
Handling sections are identical and they can be
configured:
in cold redundancy to provide high
reliability, or
in master-slave mode, so to run in
parallel thus improving the processing
capabilities
HPVN-IPM Block
HPVNDiagram
Multi Ocular Smart System
14
MOSS
HPVN
The Image Processing section is dedicated to the
intensive-computing required by the algorithms,
and it is based on the Xilinx Virtex-5XQR5VFX130,
the industry's first high performance rad-hard
reconfigurable FPGA
HPVN 3D Model
The following classes of image processing
algorithms are mainly foreseen:
Image compression based on Wavelet
transform
3D stereo vision
Feature Extraction & Tracking (FET)
The Data Handling sections adopt each one 32 bit
LEON3 SPARC processor (implemented, as soft IP
core, in a radiation tolerant anti-fuse FPGA) and
implements the control & communication
functionalities of the unit, and the high-level
image processing algorithms for visual navigation
A particular of the IPM inside HPVN
Multi Ocular Smart System
The Power Conditioning and Distribution module
is based on a fully redundant architecture that
guarantees a single fault tolerance
15
MOSS
HPVN – Main Specifications
I/O
Pixel Rate
Power Characteristics
Volume
Mass
Image Processing
Algorithms
Redundant Power Input Interface
High Priority Commands Interfaces
Redundant CAN bus Interfaces (OBDH side)
2 x SpaceWire Interfaces
Redundant CAN bus Interfaces (Optical Head side)
Multi-Ocular Camera Power Supply (Nominal + Redundant)
Multi-Ocular Camera Temperature Control Interface
CCD Camera Power Supply (Nominal + Redundant)
CDD Camera Temperature Control Interface
Video data input left (Nominal + Redundant)
Video data input right (Nominal + Redundant)
Camera external trigger left (Nominal + Redundant)
Camera external trigger right (Nominal + Redundant)
120Mpixel/s (2 x 30Mpix/s (nominal) + 2 x 30Mpix/s (redundant) or
4x30Mpixel/s)
Power input: 28V ± 25%
Communication & Control: 6.9W@28V
One active FPGA Processing section: 15.1W @28V
Two active FPGA Processing sections: 23.3W@28V
249.3 x 177 x 55.2 mm3 (excluding fixation points)
2255 g
Real-Time Stereo vision
Wavelet based Image compression @30fps, CF=(8 ÷100) in lossy o loss less
mode
Feature extraction & tracking
Multi Ocular Smart System
16