Hi i am in the process of deciding the architecture of the hardware of our next robot. We need it to interface to 1)GPS 2)implement an inertial navigation system (INS) 2.1)interface to MEMS gyro 2.2)interface to MEMS accelerometer 3)Implement a wireless LAN 802.11b 4)Drive 4 hbridges using pwm. 5)also run the control algorithm 6)maybe interface to a camera for final positioning for docking of two robots as the inertial system might not be too accurate. Please note that the above requirements mean that floating point calculations are invevitable, Could someone tell me the MIPS rating of the MSP430 processors. Can someone adivse if its wise to go in for a microcontroller like msp430...Or any microcontroller at all? I am considering using a PC104 based industrial microprocessor board , since all this i feel is just too much for a microcontroller. Please debate this with me. I shall highly appreciate you precious time and input best regards Harshit Suri __________________________________
Hardware architectural decisions
Started by ●October 28, 2003
Reply by ●October 28, 20032003-10-28
Well, a variety of processors, frame grabbers and lan interfaces are
certainly off the shelf in pc104, so if you have the space and budget (and
if you pay your engineers anything it's still much cheaper) you'd have
a
flexible and vendor-independent solution.
--Bruce
> I am considering using a PC104 based industrial
> microprocessor board , since all this i feel is just
> too much for a microcontroller. Please debate this
> with me. I shall highly appreciate you precious time
> and input
>
> best regards
> Harshit Suri
Reply by ●October 28, 20032003-10-28
You might want to take a look at http://www.acroname.com/ All though the MSP430 could work, if you don't need the low power qualities there are better processors to use. For Robotics you really need a multi-tasking capable system. Look into using Rabbit Controllers http://www.rabbitsemiconductor.com/ . They have onboard Ethernet and inexpensive development software. You might also consider a network of microcontrollers (PICs) on an I2C or 485 bus handling the housekeeping functions and a PC-104 for high level functions. The CMU cam is easy to interface to and may be a solution to your video docking problem. Bob Waters > Hi > i am in the process of deciding the architecture of > the hardware of our next robot. We need it to > interface to > 1)GPS > 2)implement an inertial navigation system (INS) > 2.1)interface to MEMS gyro > 2.2)interface to MEMS accelerometer > 3)Implement a wireless LAN 802.11b > 4)Drive 4 hbridges using pwm. > 5)also run the control algorithm > 6)maybe interface to a camera for final positioning > for docking of two robots as the inertial system might > not be too accurate. > > Please note that the above requirements mean that > floating point calculations are invevitable, > > Could someone tell me the MIPS rating of the MSP430 > processors. > > Can someone adivse if its wise to go in for a > microcontroller like msp430...Or any microcontroller > at all? > > I am considering using a PC104 based industrial > microprocessor board , since all this i feel is just > too much for a microcontroller. Please debate this > with me. I shall highly appreciate you precious time > and input > > best regards > Harshit Suri > > __________________________________ > > > > . > > > > ">http://docs.yahoo.com/info/terms/ > >
Reply by ●October 28, 20032003-10-28
harshit suri wrote: > Hi > i am in the process of deciding the architecture of > the hardware of our next robot. We need it to > interface to > 1)GPS > 2)implement an inertial navigation system (INS) > 2.1)interface to MEMS gyro > 2.2)interface to MEMS accelerometer > 3)Implement a wireless LAN 802.11b > 4)Drive 4 hbridges using pwm. > 5)also run the control algorithm > 6)maybe interface to a camera for final positioning > for docking of two robots as the inertial system might > not be too accurate. Why make life so hard for yourself? use a simple optical set up for alignment, bring it to bear wjhen the two devices are close, the INS will get you that close > > Please note that the above requirements mean that > floating point calculations are invevitable, No they don't. I run a system with dual independant INS + GPS + a pile of other sensors and devices, not a hint of a float anywhere. Floats are too inaccurate so I use simple, fast integer maths. > > Could someone tell me the MIPS rating of the MSP430 > processors. MIPs rating is meaningless. What matters most is time to execute a specific task. I will give you a real world example. The worlds first computer used for a real world task was the Collosus, used during WWII to crack the enigma code. It was rated at less than 0.2MIPS, yet about 2 years ago a team set out to reconstruct its operation. They ran this against a state of the art pentium. the Colossus cracked the code in hours, tyhe Pentium in days despite being rated at many hundreds of MIPs. Why? simply because the pentium is a general purpose computer optimised for data shuffling. The Colossus was a dedicated single task machine dedicated to cracking the enigma code. > > Can someone adivse if its wise to go in for a > microcontroller like msp430...Or any microcontroller > at all? My system runs entirely on a single MSP430, at 8MHz, however I would separate the motor controller from the rest of the stuff. I assume that you have collision detection, motor stall detection etc, edge detection, ie don't let it fall off the edge of a table. i would subordinate these basic systems to the motor control CPU, and subordinate IT to the main sensor and comms CPU. Will this be an autonomous machine or tethered (I consider R/C control to be a tether) > > I am considering using a PC104 based industrial > microprocessor board , since all this i feel is just > too much for a microcontroller. Please debate this > with me. I shall highly appreciate you precious time > and input Depends how you implement the PC104, since this is basically a data shuffling machine it may perform worse than a dedicated bit munching microcontroller. I'd use Pc104 when the majority of the design was data collection user interface etc, not when it was real world interfacing and sensor processing. Al
Reply by ●October 28, 20032003-10-28
--- In msp430@msp4..., "Bob Waters" <b_waters@c...> wrote: > All though the MSP430 could work, if you don't need the low power qualities > there are better processors to use. > For Robotics you really need a multi-tasking capable system. Just to pick nits, the MSP430 is eminently well-suited to multitasking ... --Andrew Kalman aek ... at ... pumpkin ... inc ... dot ... com
Reply by ●October 29, 20032003-10-29
Hi Everyone thankyou all for your replies. Al thankyou for your advice. Could you please give me some links / documents for your INS + GPS implementation this is an indoor robot. autonomous and it hovers on top of a air hockey table using the principle of air bearings i am using an air hockey table to simulate the frcitionless environment of space in 2Dimensions. I really dont know if ill be able to use gps as its in door but its my first draft of design so it might change. could you please share any design docs /implementation of your INS system have u interfaced MEMS gyros + MEMS accelerometer to the MSP430 ? the objective of the robots are to cooperate to build a structure.so they fly around catch onto things and then cooperate to construct structures :-) once again thanks for your help everyone cheers Harshit Suri __________________________________
Reply by ●October 29, 20032003-10-29
Hi harshit, I thought this might be relevant, I am not sure if technology has improved significantly in this area, but it has been my experience that GPS is not accurate beyond one meter or so. When I used it, MILITARY, it was only accurate to ten meters, but I have heard it improved, just not sure to much better than a meter. Nick -----Original Message----- From: harshit suri [mailto:suri_list@suri...] Sent: Tuesday, October 28, 2003 10:36 PM To: msp430@msp4... Subject: Re: [msp430] Hardware architectural decisions Hi Everyone thankyou all for your replies. Al thankyou for your advice. Could you please give me some links / documents for your INS + GPS implementation this is an indoor robot. autonomous and it hovers on top of a air hockey table using the principle of air bearings i am using an air hockey table to simulate the frcitionless environment of space in 2Dimensions. I really dont know if ill be able to use gps as its in door but its my first draft of design so it might change. could you please share any design docs /implementation of your INS system have u interfaced MEMS gyros + MEMS accelerometer to the MSP430 ? the objective of the robots are to cooperate to build a structure.so they fly around catch onto things and then cooperate to construct structures :-) once again thanks for your help everyone cheers Harshit Suri __________________________________ <http://rd.yahoo.com/M4081.4074964.5287182.1261774/D=egroupweb/S050053 78:HM/A32161/R=0/SIGp5b9ris/*http:/www.ediets.com/start.cfm?code0509 &media=atkins> click here <http://us.adserver.yahoo.com/l?M4081.4074964.5287182.1261774/D=egroupmai l/S=:HM/A32161/rand(9295915> . ">http://docs.yahoo.com/info/terms/> Terms of Service.
Reply by ●October 29, 20032003-10-29
----- Original Message ----- From: "harshit suri" <suri_list@suri...> To: <msp430@msp4...> Sent: Tuesday, October 28, 2003 11:16 PM Subject: [msp430] Hardware architectural decisions > Hi > i am in the process of deciding the architecture of > the hardware of our next robot. We need it to > interface to > 1)GPS > 2)implement an inertial navigation system (INS) > 2.1)interface to MEMS gyro > 2.2)interface to MEMS accelerometer > 3)Implement a wireless LAN 802.11b > 4)Drive 4 hbridges using pwm. > 5)also run the control algorithm > 6)maybe interface to a camera for final positioning > for docking of two robots as the inertial system might > not be too accurate. > > Please note that the above requirements mean that > floating point calculations are invevitable, > > Could someone tell me the MIPS rating of the MSP430 > processors. > > Can someone adivse if its wise to go in for a > microcontroller like msp430...Or any microcontroller > at all? I don't think the MSP430 will cope with that lot. Something like the new Philips ARM probably will, though. Have a look at the spec: http://www.semiconductors.philips.com/pip/LPC2104.html I've got details of a simple dev. system using it on my web site. Leon -- Leon Heller, G1HSM Email: aqzf13@aqzf... http://webspace.webring.com/people/jl/leon_heller/
Reply by ●October 29, 20032003-10-29
Hi Harshit, If you need to receive GPS signals indoors, there are GPS re-radiating kits available. We purchased this one and it works great. There a little pricey, but it sure beats having to take everything outside to do your testing! http://www.navtechgps.com/supply/rerad.asp HTH, Greg ----- Original Message ----- From: "harshit suri" <suri_list@suri...> To: <msp430@msp4...> Sent: Tuesday, October 28, 2003 10:36 PM Subject: Re: [msp430] Hardware architectural decisions > Hi Everyone > > thankyou all for your replies. Al thankyou for your > advice. Could you please give me some links / > documents for your INS + GPS implementation > this is an indoor robot. autonomous and it hovers on > top of a air hockey table using the principle of air > bearings > i am using an air hockey table to simulate the > frcitionless environment of space in 2Dimensions. > I really dont know if ill be able to use gps as its in > door but its my first draft of design so it might > change. could you please share any design docs > /implementation of your INS system > have u interfaced MEMS gyros + MEMS accelerometer to > the MSP430 ? > > the objective of the robots are to cooperate to build > a structure.so they fly around catch onto things and > then cooperate to construct structures :-) > > once again thanks for your help everyone > cheers > Harshit Suri > > __________________________________ > > > > . > > > > ">http://docs.yahoo.com/info/terms/ > >
Reply by ●October 29, 20032003-10-29
Hi Harshit, as you can see from the private post I can't really tell
you
much about my system. I do agree with leon regarding the ARM. I know
leon has a simple eval board for it, but I'm still waiting for a
complete tool set with debuggers etc at a price I can afford. GPS
98indoors is a problem without a relay system. this would be expensive.
rather than that I would use INS plus an optical system, since your
total movement is within a very confined area. I would grid this so
that, say, each grid was 5cm on a side. Use magnetometers rather than
gyros, since you really only need to know your compass heading. rate of
rotation can be easily calculated from this. Look at the Honeywell
HMC1052. I use the HMC1053 3D part, and they are excellent. Add to this
just a single accelerometer. This gives you motion relative to the
direction of heading. Add a second accelerometer in the Z axis if tilt
is important. Honey well have excellent notes on very accurate tilt
compensated compasses. Next I would add a few lasers in a simple matrix
to the table. The narrow beam is required. stick a few suitable optical
sensors around the base of the robot. modulate each laser at a different
frequency. (you don't need lots of them) And simply use timer capture to
determine which beam is detected where. this will allow to to
periodically correct your absolute position.
I know you are trying to simulate 2D motion in space, and this doesn't
quite do it. However conventional GPS wouldn't work in space anyway, off
the shelf GPS models the earths atmosphere to correct for its effects on
the signals. (or they did in 1993 when I last worked on the internal
algorithms) What this does is give you some intimate experience in the
complexities of tracking. For space I would use gyros, for rate of
rotation, accelerometers for motion and lasers still in place of GPS for
periodic correction. It would be far easier to align two moving objects
with a cooperative optical system using relative position than to try
and match them with absolute INS based position.
Al
harshit suri wrote:
> Hi Everyone
>
> thankyou all for your replies. Al thankyou for your
> advice. Could you please give me some links /
> documents for your INS + GPS implementation
> this is an indoor robot. autonomous and it hovers on
> top of a air hockey table using the principle of air
> bearings
> i am using an air hockey table to simulate the
> frcitionless environment of space in 2Dimensions.
> I really dont know if ill be able to use gps as its in
> door but its my first draft of design so it might
> change. could you please share any design docs
> /implementation of your INS system
> have u interfaced MEMS gyros + MEMS accelerometer to
> the MSP430 ?
>
> the objective of the robots are to cooperate to build
> a structure.so they fly around catch onto things and
> then cooperate to construct structures :-)
>
> once again thanks for your help everyone
> cheers
> Harshit Suri
>
> __________________________________
>
>
>
> .
>
>
>
> ">http://docs.yahoo.com/info/terms/
>
>
>