Following table shows the data rate requirements
--------------------------------------------------------------------------------------------
Item Data rate Data rate with Comments Net Data Rate
Compression
--------------------------------------------------------------------------------------------
Uplink
Rover Status 20 int at 4 Hz 1280 bps Continuous
2 Mbps
Panospheric Cam 2k x 2k @10 bit, 1.88 Mbps (at Continuous
era 6-8 Hz 170:1 compres
= 32 Mbps sion)
Front Camera 1k X 1k @ 8 bit Occasionally
Back Camera 1k X 1k @ 8 bit Occasionally
Stereo Cameras
Downlink
Command & Con 20 int + 20 floats 3840 bps Continuous
trol at 4 Hz. 500 Kbps avail
able
Camera Control Which camera, Com
pression Ratio, FOV
Downloading code - - Occasional
--------------------------------------------------------------------------------------------
We looked at various options to meet these requirements using omnidirectional antennas and it became clear that directional antennas are required. This adds challenge of keeping the antenna on the mobile robot aligned to the antenna on the control station. We couldn't find any off-the-shelf device that would achieve the pointing requirements needed and so we have to design and build an antenna pointing device ourself. The detailed design, budgets and the schedule for communication system, including the pointing device are described in the following sections.
The main components of the communication system are:
Rover End

The high bandwidth (BR2040 from Aironet, 2 Mbps) ethernet bridge is to support the imagery requirements. There are three high resolution cameras (Panospheric camera, front camera and the back camera) mounted on the robot. The three cameras are attached to a 3Way Mux which is controlled via an RS232 connection to the imaging computing. This Mux determines which of the cameras will be providing the image that is framegrabbed and which camera is receiving the control signals. The output of the Mux goes to the framegrabber board which places the image into the memory of the pc. This is then compressed and sent to real-time computer over the ethernet and then to the wireless ethernet bridge. This bridge communicates with the bridge on the relay station using high gain antennas and pointing mechanism. Pointing Mechanism is discussed in detail in following sections.
The Low bandwidth radio (115 Kbps) is used for status/command/and control. This radio operates on lower frequency (900 MHz) compared to the ethernet bridge and can take some Non-LOS. Since it has much lower bandwidth it doesn't need high gain antennas either. Thus it is more reliable than the bridge and so is used for crucial status/command and control. It will also transmit low res imagery from a "QuickCam" so that the rover can be teleoperated even in the absence of imagery from Panospheric camera.
Relay station/Satellite Link
The relay station has similar bridge and radio to communicate to the rover.
Control Truck
The ethernet and serial are read in by computer and converted to a single ethernet stream. This computer also acts a local control station. The ethernet is converted to RS 449 using a router (SYSCO 2511 or SYSCO 2501) which is fed to the satellite transponder.
Control Station
The satellite downlink is in Virginia (or Florida). From there it is send to Pittsburgh Science Center using land lined. On the stateside, the compressed image is received from the landline and sent to the decompression computing.
A detailed schematic showing connectivity of various sites is shown below:
The Serial (RF modem) link to Nomad should use a PPP or SLIP protocol. This will provide the guaranteed end-to-end transport layer needed for the robot telemetry and operator commands. Thus a PPP driver for the Real Time 68030 will be needed(1). The Real Time computer will have two IP addresses; one to handle the PPP data coming over the serial line, and one to enable addressing on Nomad's onboard ethernet. It might be prudent to devote the 68030 board to handling these two IP connections. We'll have to see how that influences the laser scanner processing as well. Relay Station GPS data, which must be made available to Nomad to enable differential GPS computations, can be communicated over the RF modem link via a multiplexer. This will allow the data to be communicated without further burdening the Real Time computer, though it will require the addition of a Mux at the relay station and one onboard Nomad. The Arlan bridge at the relay station will effectively isolate any bad packets resulting from poor high speed communication with Nomad. Thus it is ok to have the Arlan on the same ethernet as the rest of the Command Truck local network. We will use a router (CISCO 2511)(2) to combine ethernet and serial link and convert it to RS449 (or V.35) serial that goes into satellite modem. A router would be used at the Pittsburgh end to recover the serial and ethernet steram. By simply adding a modem to the stateside network, we will be able to monitor and debug networking problems during the mission.
Panospheric data will be broadcast using IP Multicasting. That is, the panospheric compression box will send out compressed data with a multicast destination address. Freeware code for this is available in the VAT (Internet Video and Audio Teleconferencing software) toolkit, available from Lawrence Livermore. This data will be available in the Command Truck, and shipped stateside via an IP tunnel in the Cisco router on the Command Truck.
The current configuration does not include an Internet link. However, such a link could be easily added to the stateside network via a gateway machine. We must make certain to keep Internet packets off of our local network, however, mainly because we expect the panospheric data to occupy much of the local network's bandwidth. And just by connecting to the Internet, bridge and routing table usage will explode (CMU's local net alone has several thousand nodes that will all try to register themselves).
Antenna pointing is one of the most critical component of the communication system. To maintain the communication at the required data rate we need to keep the high gain antennas (on the rover and the relay station) pointed towards each other.
The pointing requirements are given in Table 2.
------------------------------------------------------ Item Value Comments ------------------------------------------------------ Elevation range -60 to + 60 deg Azimuth range 360 deg Continuous Elevation Rate 60 deg/sec Azimuth Rate 30 deg/sec Elevation Acceleration 190 deg/sec^2 Azimuth Acceleration 190 deg/sec^2 Pointing Accuracy 2.5 deg Aggregate Pay Load Mass 2 kg Payload Size 27.3 x 34.0 x 2 cm Peak Torque Stall Torque Velocity Smoothness Stability Environmental ------------------------------------------------------

Following sensors are used for antenna pointing: DGPS, encoders, compass, Inclinometer. IMU, if available can be used to improve the performance.
DGPS, Compass, Inclinometers and Encoders are minimum sensors required for pointing control and are baseline sensors for the design. But they achieve only open loop pointing. There is no way to find the actual pointing error or evaluate the quality of pointing using these sensors. It would be useful (but not necesary) to have a feedback method. We are looking at several options to close the feedback loop, including:
Given the references of the Antenna Positioning Solver sw module, it will point the antenna with two position PID loops and trajectory generators that control a velocity command to the pointing amps. A hardware board will read incremental encoders from the feedback part of the motors. This module will take care of the maximum azimuth angles related to wire winding (this last applicable only if there are no slip rings).
-----------------------------------------------------------------------------------------------------------
Item Unit Price Quantity Total Total w/ Comments
margin
-----------------------------------------------------------------------------------------------------------
ROVER
Ethernet bridge 3000 1 3000 3300 BR2040-Aironet
16 dB Antenna 424.14 1 424.14 424.14 Ordered
Radio Modem 2000 1 2000 2200 Freewave/Pacific
Crest
Cables/connectors 1000 1 1000 1100
6424.14 7024.24
Antenna Pointing Device
Mechanism 2000 1 2000 2200
Slip Rings 800 1 800 900
Pan motor 2000 1 2000 2300
Tilt Motor 1200 1 1200 1400
DGPS + radio 5000+ 2000 1 7000 8000 Novel/ Trimble
GyroCompass 2375 1 2375 2375 KVH Industries
Feedback Sensing 2000 1 2000 2400 Laser tracking
Encoders 5000 2 0 0 Donation- BEI
Cables/connectors 1000 - 1000 1100
18,375 20, 675
RELAY STATION
Ethernet Bridge 3000 1 3000 3300 BR2040- Aironet
High gain antenna 427.14 1 427.14 427.14 Ordered; H+S
1324.19.0010
Radio Modem 2000 1 2000 2200 Freewave/Pacific
Crest
DGPS + radio 5000+2000 1 7000 8000 Novel/ Trimble
Antenna Pointing 1000 1 1000 1200
Cables/connectors 1000 1 1000 1100
14427.14 16,227.14
SATELLITE LINK
Link Time 28,100 1 month 28,100 30,000 Lyman Broth
ers, Fruit
Heights, UT
Referred by
NASA Ames.
Equipment rental 12,500 2 month 25,000 25,000
landline 5000 1 month 5000 5500
Router 2000 1 2000 2500
Consulting(a) 7000 8000
Shipping and Han 9000 2 18,000 18,000
dling
85,100 89,000
OTHER
8 dB Antenna 65.00 2 0.00 0.00 Donation
RS 422-232 Con 69.95 1 69.95 69.95 Gyro-compass
verter testing
Laptop 4300 1 4300 4300 Ordered
Router 2000 1 2000 2500 Control Station
Cables 1000 1 1000 1100
7369.95 7969.95
TOTAL 131,696.23 140,896.23
Satellite Equipment- Shipping - 9000 Possible Saving
If Ratler radio works for range... - 4500 Good Chance
Laptop - 4300 Other Budget
113,896.23(b)
-----------------------------------------------------------------------------------------------------------
Issues:
The problem is presented as follows: given the relative position of the rover and the repeater station, and given the angles of the roiver respect to the general frame (earth frame), what are the values of pitch and roll necesaries to point the rover's antenna to the repeater station.
Notation
Robot pitch
Robot yaw
Robot Azimuth
Azimuth motor encoder
Elevation motor encoder
Coordinates of the relay station (GPS)
Coordinates of the robot (GPS)
Location of the pointing frame w.r.t the vehicle frame
This frame has its x-axis pointing towards the East. The y-axis is pointing towards the North. The z-axis is pointing upwards in the local vertical. The origin of this frame is somewhere in the earth, and it is irrelevant because we are interested in the relative position of the rover and the station.
There is another {V} frame attached to the rover (vehicle) in such a way that the origin coincides with the Digital Gyro Compass. The y-axis points forward. The x-axis points to the right side of the vehicle. Finally, the z-axis points up in the local vertical. We know the relative position of {V} with respect to the {G} using the GPS output. The DGC gives the yaw, pitch and roll of the vehicle.
There is a mast on the vehicle, and it has the antenna to be pointed on the top of it. The antenna has been designed to have two axis controlled, that is the yaw and pitch. The frame mounted on the antenna {A} has the y-axis pointing toward the electrical axis of the antenna. So the problem is to control the antenna to intersect the projection of the y-axis of {B} with the station.
There is also an auxiliar frame mounted on the antenna axis intersection, that will be very useful as we will see. That is a frame with its origin coincident with the {A} origin, but with the axis parallel to the {V} axis.
Tha goal or station vector is S, and it is also the result of GPS data.
Now, once S is expressed in {B} cartesian coordinates, it happens that the antenna moves in spherical coordinates. Then, if we use the cartesian to spherical coordinates, we will get the angles we need to point rthe antenna to. This is the inverse kinematics part because given the point, we need the joint angles.
Once in {V} coordinates, the translation to the top of the mast is accomplished by multiplyig the vector by the next transformation matrix:
Now we have that the full transformation matrix from the {G} frame to the {B} frame is the product of:
Multiplying this matrix and the station vector, we obtain the vector in {B} coordinates. Now, to calculate the yaw angle of the antenna, we note that that angle is the angle of the proyection of the station vector on the BX-BY plane, with respect to the BY axis.
Similarly, the pitch is the angle between the S vector and the BX-BY plane. The equation's to solve for those angles are:
It is possible to obtain equations for the coordinates of the station in the {B} frame. After multiplying the matrices and the vector, we get the following equations: