Hi, i am starting entering on the fixed wing VANT reasearch field, and i was browsing the options for the control system architecture. I could find two possible ways i could go. A TMS570 microcontroller + Some safe RTOS (SafeRTOS, =ECC/OS-II...), or a PC104 (with non defined processor architecture) + QNX. In most of the cases the last is the one that got much more horsepower, but still is heavy, bulky and will probably have a worst RT response than the first one. However the first one got a very limited processing power (180Mhz frequency), so am not sure if that is going to be enough for now. I belive t will, as i will have a separare processor for the AHRS system. Any comments or sugestions regarding this architecture? Any one with know-how on the field? Thank you!
VANT processor/RTOS choice. Need sugestions
Started by ●February 26, 2012
Reply by ●February 26, 20122012-02-26
On Sun, 26 Feb 2012 13:44:59 -0800, Sink0 wrote:> Hi, i am starting entering on the fixed wing VANT reasearch field, and i > was browsing the options for the control system architecture. I could > find two possible ways i could go. A TMS570 microcontroller + Some safe > RTOS (SafeRTOS, μC/OS-II...), or a PC104 (with non defined processor > architecture) + QNX. In most of the cases the last is the one that got > much more horsepower, but still is heavy, bulky and will probably have a > worst RT response than the first one. However the first one got a very > limited processing power (180Mhz frequency), so am not sure if that is > going to be enough for now. I belive t will, as i will have a separare > processor for the AHRS system. > > Any comments or sugestions regarding this architecture? Any one with > know-how on the field? > > Thank you!So, what the heck does "VANT" stand for? Who's limiting your processor choices? It would seem that if you can use any old PC-104 board, that you can use any old processor at all. The real-time response is going to mostly be a function of the operating system (and possibly the peripherals), and as far as I know QNX has decent RT chops -- why do you believe it'll be slower to respond? -- My liberal friends think I'm a conservative kook. My conservative friends think I'm a liberal kook. Why am I not happy that they have found common ground? Tim Wescott, Communications, Control, Circuits & Software http://www.wescottdesign.com
Reply by ●February 26, 20122012-02-26
On Feb 26, 7:09=C2=A0pm, Tim Wescott <t...@seemywebsite.com> wrote:> On Sun, 26 Feb 2012 13:44:59 -0800, Sink0 wrote: > > Hi, i am starting entering on the fixed wing VANT reasearch field, and =i> > was browsing the options for the control system architecture. I could > > find two possible ways i could go. =C2=A0A TMS570 microcontroller + Som=e safe> > RTOS (SafeRTOS, =CE=BCC/OS-II...), or a PC104 (with non defined process=or> > architecture) + QNX. In most of the cases the last is the one that got > > much more horsepower, but still is heavy, bulky and will probably have =a> > worst RT response than the first one. However the first one got a very > > limited processing power (180Mhz frequency), so am not sure if that is > > going to be enough for now. I belive t will, as i will have a separare > > processor for the AHRS system. > > > Any comments or sugestions regarding this architecture? Any one with > > know-how on the field? > > > Thank you! > > So, what the heck does "VANT" stand for? > > Who's limiting your processor choices? =C2=A0It would seem that if you ca=n use> any old PC-104 board, that you can use any old processor at all. > > The real-time response is going to mostly be a function of the operating > system (and possibly the peripherals), and as far as I know QNX has > decent RT chops -- why do you believe it'll be slower to respond? > > -- > My liberal friends think I'm a conservative kook. > My conservative friends think I'm a liberal kook. > Why am I not happy that they have found common ground? > > Tim Wescott, Communications, Control, Circuits & Softwarehttp://www.wesco=ttdesign.com Sorry, UAVs (VANT is the word in portuguese).. Well i belive QNX got a good enough response (as it is widely used on the field). Still it needs a much more complex computer system, that is going to be bulky, heavy, and much more power hungry. And TMS570 got several redundancies that makes it nice for safe-crytical applications.
Reply by ●February 27, 20122012-02-27
On 27/02/2012 00:11, Sink0 wrote:> On Feb 26, 7:09 pm, Tim Wescott<t...@seemywebsite.com> wrote: >> On Sun, 26 Feb 2012 13:44:59 -0800, Sink0 wrote: >>> Hi, i am starting entering on the fixed wing VANT reasearch field, and i >>> was browsing the options for the control system architecture. I could >>> find two possible ways i could go. A TMS570 microcontroller + Some safe >>> RTOS (SafeRTOS, μC/OS-II...), or a PC104 (with non defined processor >>> architecture) + QNX. In most of the cases the last is the one that got >>> much more horsepower, but still is heavy, bulky and will probably have a >>> worst RT response than the first one. However the first one got a very >>> limited processing power (180Mhz frequency), so am not sure if that is >>> going to be enough for now. I belive t will, as i will have a separare >>> processor for the AHRS system. >> >>> Any comments or sugestions regarding this architecture? Any one with >>> know-how on the field? >> >>> Thank you! >> >> So, what the heck does "VANT" stand for? >> >> Who's limiting your processor choices? It would seem that if you can use >> any old PC-104 board, that you can use any old processor at all. >> >> The real-time response is going to mostly be a function of the operating >> system (and possibly the peripherals), and as far as I know QNX has >> decent RT chops -- why do you believe it'll be slower to respond? >> >> -- >> My liberal friends think I'm a conservative kook. >> My conservative friends think I'm a liberal kook. >> Why am I not happy that they have found common ground? >> >> Tim Wescott, Communications, Control, Circuits& Softwarehttp://www.wescottdesign.com > > Sorry, UAVs (VANT is the word in portuguese)..Clear as mud. Try assuming that we know all about embedded development, but nothing about your application - /then/ try to ask a question. "UAV" and "AHRS" may be obvious to you - but I could probably think of a dozen different things "UAV" could stand for. The only thing I can think of off-hand that fits with the "fixed wing" description, is "unmanned aviation vehicle". I really hope your work is firmly in the "research" area - you don't want to consider leaving the ground while you are asking the kind of questions you are asking now. Of course, it is much better to be inexperienced and asking questions, than inexperienced and /not/ asking questions!> > Well i belive QNX got a good enough response (as it is widely used on > the field). Still it needs a much more complex computer system, that > is going to be bulky, heavy, and much more power hungry. And TMS570 > got several redundancies that makes it nice for safe-crytical > applications.Safety is about reducing the risk of faults developing, reducing the risk of those faults developing into failures, and reducing the consequence of those failures. Identifying and strengthening the weakest links is vital. If you have a system where the requirements justify using a redundant cpu microcontroller like the TMS570 (or one from Freescale - they have a wider selection), then there is no way that a PC-qualified processor on a PC-style PC104 motherboard will ever be good enough. It is possible that a PC card running QNX could be /part/ of the solution, but not a critical part (it could handle things like communication, or camera control, while the actual flying is left to the TMS570).
Reply by ●February 28, 20122012-02-28
> > Clear as mud. > > Try assuming that we know all about embedded development, but nothing > about your application - /then/ try to ask a question. =A0"UAV" and "AHRS="> may be obvious to you - but I could probably think of a dozen different > things "UAV" could stand for.Sorry, lots of mistakes here. Yea you are right, here UAV stands for Unmanned aerial vehicle.> The only thing I can think of off-hand that fits with the "fixed wing" > description, is "unmanned aviation vehicle". =A0I really hope your work i=s> firmly in the "research" area - you don't want to consider leaving the > ground while you are asking the kind of questions you are asking now. > Of course, it is much better to be inexperienced and asking questions, > than inexperienced and /not/ asking questions! >Haha, for sure we are not considering leaving ground before we are much more mature on the control system. The airplaine design itself is a well solved problem and we got a team with lots of experience working on that. But anyway thats why i am asking here. Actually i had a long meeting today with a guy that got much more experience and already worked with a UAV system today.> > > Well i belive QNX got a good enough response (as it is widely used on > > the field). Still it needs a much more complex computer system, that > > is going to be bulky, heavy, and much more power hungry. And TMS570 > > got several redundancies that makes it nice for safe-crytical > > applications. > > Safety is about reducing the risk of faults developing, reducing the > risk of those faults developing into failures, and reducing the > consequence of those failures. =A0Identifying and strengthening the > weakest links is vital. > > If you have a system where the requirements justify using a redundant > cpu microcontroller like the TMS570 (or one from Freescale - they have a > wider selection), then there is no way that a PC-qualified processor on > a PC-style PC104 motherboard will ever be good enough. =A0It is possible > that a PC card running QNX could be /part/ of the solution, but not a > critical part (it could handle things like communication, or camera > control, while the actual flying is left to the TMS570).Thats my whole point on asking here still several system go to a standard PC with QNX choice but on the meeting i got my answer why. The whole point is about certification and developing effort. However i still like more the idea on working with TMS570/PX30 on the low level control system. Now i got a second question regarding on that. Any one knows if uc/OS- II or SafeRTOS are certifiable on DO178B class-A. Thank you! Any comments and sugestions are wellcome!
Reply by ●February 28, 20122012-02-28
On Tue, 28 Feb 2012 10:05:34 -0800, Sink0 wrote:>> Clear as mud. >> >> Try assuming that we know all about embedded development, but nothing >> about your application - /then/ try to ask a question. "UAV" and >> "AHRS" may be obvious to you - but I could probably think of a dozen >> different things "UAV" could stand for. > > > Sorry, lots of mistakes here. Yea you are right, here UAV stands for > Unmanned aerial vehicle. > > >> The only thing I can think of off-hand that fits with the "fixed wing" >> description, is "unmanned aviation vehicle". I really hope your work >> is firmly in the "research" area - you don't want to consider leaving >> the ground while you are asking the kind of questions you are asking >> now. Of course, it is much better to be inexperienced and asking >> questions, than inexperienced and /not/ asking questions! >> > Haha, for sure we are not considering leaving ground before we are much > more mature on the control system. The airplaine design itself is a well > solved problem and we got a team with lots of experience working on > that. > > But anyway thats why i am asking here. > > Actually i had a long meeting today with a guy that got much more > experience and already worked with a UAV system today. > > >> > Well i belive QNX got a good enough response (as it is widely used on >> > the field). Still it needs a much more complex computer system, that >> > is going to be bulky, heavy, and much more power hungry. And TMS570 >> > got several redundancies that makes it nice for safe-crytical >> > applications. >> >> Safety is about reducing the risk of faults developing, reducing the >> risk of those faults developing into failures, and reducing the >> consequence of those failures. Identifying and strengthening the >> weakest links is vital. >> >> If you have a system where the requirements justify using a redundant >> cpu microcontroller like the TMS570 (or one from Freescale - they have >> a wider selection), then there is no way that a PC-qualified processor >> on a PC-style PC104 motherboard will ever be good enough. It is >> possible that a PC card running QNX could be /part/ of the solution, >> but not a critical part (it could handle things like communication, or >> camera control, while the actual flying is left to the TMS570). > > Thats my whole point on asking here still several system go to a > standard PC with QNX choice but on the meeting i got my answer why. > > The whole point is about certification and developing effort. However i > still like more the idea on working with TMS570/PX30 on the low level > control system. > > Now i got a second question regarding on that. Any one knows if uc/OS- > II or SafeRTOS are certifiable on DO178B class-A.I know that uC/OS-II mentions certification on their web site -- you might want to check. Does your UAV need to be certifiable to DO178B class A? I would have thought that certification was just for the (nonexistent) people on board -- is DOT also then concerned about bystanders? Even in this day and age I'd be tempted to partition the design and put the class A bits on a separate processor without a 3rd-party OS (and where are you gonna get a certified compiler?). But that's me, and I guess it depends strongly on just how much functionality you need in the class A part. -- My liberal friends think I'm a conservative kook. My conservative friends think I'm a liberal kook. Why am I not happy that they have found common ground? Tim Wescott, Communications, Control, Circuits & Software http://www.wescottdesign.com
Reply by ●February 28, 20122012-02-28
On 28/02/12 21:49, Tim Wescott wrote:> On Tue, 28 Feb 2012 10:05:34 -0800, Sink0 wrote: > > >>> Clear as mud. >>> >>> Try assuming that we know all about embedded development, but nothing >>> about your application - /then/ try to ask a question. "UAV" and >>> "AHRS" may be obvious to you - but I could probably think of a dozen >>> different things "UAV" could stand for. >> >> >> Sorry, lots of mistakes here. Yea you are right, here UAV stands for >> Unmanned aerial vehicle. >> >> >>> The only thing I can think of off-hand that fits with the "fixed wing" >>> description, is "unmanned aviation vehicle". I really hope your work >>> is firmly in the "research" area - you don't want to consider leaving >>> the ground while you are asking the kind of questions you are asking >>> now. Of course, it is much better to be inexperienced and asking >>> questions, than inexperienced and /not/ asking questions! >>> >> Haha, for sure we are not considering leaving ground before we are much >> more mature on the control system. The airplaine design itself is a well >> solved problem and we got a team with lots of experience working on >> that. >> >> But anyway thats why i am asking here. >> >> Actually i had a long meeting today with a guy that got much more >> experience and already worked with a UAV system today. >> >> >>>> Well i belive QNX got a good enough response (as it is widely used on >>>> the field). Still it needs a much more complex computer system, that >>>> is going to be bulky, heavy, and much more power hungry. And TMS570 >>>> got several redundancies that makes it nice for safe-crytical >>>> applications. >>> >>> Safety is about reducing the risk of faults developing, reducing the >>> risk of those faults developing into failures, and reducing the >>> consequence of those failures. Identifying and strengthening the >>> weakest links is vital. >>> >>> If you have a system where the requirements justify using a redundant >>> cpu microcontroller like the TMS570 (or one from Freescale - they have >>> a wider selection), then there is no way that a PC-qualified processor >>> on a PC-style PC104 motherboard will ever be good enough. It is >>> possible that a PC card running QNX could be /part/ of the solution, >>> but not a critical part (it could handle things like communication, or >>> camera control, while the actual flying is left to the TMS570). >> >> Thats my whole point on asking here still several system go to a >> standard PC with QNX choice but on the meeting i got my answer why. >> >> The whole point is about certification and developing effort. However i >> still like more the idea on working with TMS570/PX30 on the low level >> control system. >> >> Now i got a second question regarding on that. Any one knows if uc/OS- >> II or SafeRTOS are certifiable on DO178B class-A. > > I know that uC/OS-II mentions certification on their web site -- you > might want to check. >SafeRTOS certainly has lots of certifications - but again you'd want to see their website. SafeRTOS has the advantage that you can do your research and development using the free FreeRTOS version - and once you've come far enough to be fixed on the system and want to get certified, you can switch to SafeRTOS with exactly the same application code.> Does your UAV need to be certifiable to DO178B class A? I would have > thought that certification was just for the (nonexistent) people on board > -- is DOT also then concerned about bystanders? >I'd imagine that an out-of-control UAV is considered a safety hazard for people and property, and therefore needs certification.> Even in this day and age I'd be tempted to partition the design and put > the class A bits on a separate processor without a 3rd-party OS (and > where are you gonna get a certified compiler?). But that's me, and I > guess it depends strongly on just how much functionality you need in the > class A part. >I too would aim to divide up the system - regardless of any certification requirements, you'd want your most critical parts to be as simple as possible (hardware and software).
Reply by ●February 29, 20122012-02-29
> > SafeRTOS certainly has lots of certifications - but again you'd want to > see their website. =A0SafeRTOS has the advantage that you can do your > research and development using the free FreeRTOS version - and once > you've come far enough to be fixed on the system and want to get > certified, you can switch to SafeRTOS with exactly the same application > code. >That is one of the reasons i was pendng to SafeRTOS, however, they mention to be IEC61508 SIL3 certifiable, but nothing related to DO-178B. uc/OS-II does mention DO-178B.> > I'd imagine that an out-of-control UAV is considered a safety hazard for > people and property, and therefore needs certification. > >Thats is right. I am not sure how it does works on other countries, but here, after a specific weight and power it must follow the same rules as any standard airplaine. I am not from US, but the rules are inspired on USA ones.> I too would aim to divide up the system - regardless of any > certification requirements, you'd want your most critical parts to be as > simple as possible (hardware and software).Yea, that what i commented before. I will have two separate processors, one for low level control and other for mission planning. About using or not an OS, that is going to be decided on another cycle of the project. For now we are colection as much information as possible so we can start creating our development strategy. Thank you for all the help. Any new information and sugestions are more than welcome.
Reply by ●February 29, 20122012-02-29
On 29/02/2012 14:04, Sink0 wrote:>> >> SafeRTOS certainly has lots of certifications - but again you'd want to >> see their website. SafeRTOS has the advantage that you can do your >> research and development using the free FreeRTOS version - and once >> you've come far enough to be fixed on the system and want to get >> certified, you can switch to SafeRTOS with exactly the same application >> code. >> > > That is one of the reasons i was pendng to SafeRTOS, however, they > mention to be IEC61508 SIL3 certifiable, but nothing related to > DO-178B. uc/OS-II does mention DO-178B.Ask Richard Barry (the guy behind SafeRTOS and FreeRTOS) - he will be happy to help you. He hangs around this newsgroup, so he'll answer if he spots this thread. If not, his contact information is on the website.
