Microsoft Robotics Developer Studio
Microsoft Robotics Developer Studio
by Spiked3
Wed Feb 25, 2009 11:06 pm
I mentioned it in my hello thread, and was wondering has anyone looked into using Microsoft robotics studio and the robonova, or are already in the process of developing with it?
I get the impression it is a framework that is well suited for robotics. Some code would have to be written on the RN1 side, possibly in robobasic, and some 'services' written on the PC side as well. I would assume a blue tooth serial device would provide the communication transport.
Maybe I'm way off basis I don't know, but would be very interested in hearing from anyone who has looked at it more than I have and what their thoughts on it are.
I get the impression it is a framework that is well suited for robotics. Some code would have to be written on the RN1 side, possibly in robobasic, and some 'services' written on the PC side as well. I would assume a blue tooth serial device would provide the communication transport.
Maybe I'm way off basis I don't know, but would be very interested in hearing from anyone who has looked at it more than I have and what their thoughts on it are.
by i-Bot
Fri Feb 27, 2009 11:56 am
The value of MSRS would seem to depend on your skills and objectives.
Unless you are already accomplished in .net and C# then the MSRS learning curve is pretty steep.
If your objective is to package together standard Robonova moves into a more complex routine then there does seem some value to using MSRS. However MSRS does not seem to be so easily applied to the creation or execution of motions. It seems best applied to robots which are already physically stable and have an already defined set of moves. Maybe if you are looking at a higher level application like football or firefighting, then there may be value to MSRS.
Most of my time on bipeds like the Robonova is spent in creating and executing moves, and even the modelling capability of MSRS does not seem to help very much here either.
Another thing to note is that the Robonova is not very MSRS friendly in communication either. The serial port are the only available connections and these work at an effective rate of about 1200 bps and only one byte at a time. Because there is no buffer in the Robonova and often considerable delay in execution in the robonova, plus the latency of single byte transactions on the PC, then communication performance is poor. Again this means that with standard RoboNova code and Robobasic, dynamic control of servos or sensors is not practical, since you will only get a few bytes transferred per second.
Still let us know how you decide. We can help with information on the Robonova side protocols and programming.
Unless you are already accomplished in .net and C# then the MSRS learning curve is pretty steep.
If your objective is to package together standard Robonova moves into a more complex routine then there does seem some value to using MSRS. However MSRS does not seem to be so easily applied to the creation or execution of motions. It seems best applied to robots which are already physically stable and have an already defined set of moves. Maybe if you are looking at a higher level application like football or firefighting, then there may be value to MSRS.
Most of my time on bipeds like the Robonova is spent in creating and executing moves, and even the modelling capability of MSRS does not seem to help very much here either.
Another thing to note is that the Robonova is not very MSRS friendly in communication either. The serial port are the only available connections and these work at an effective rate of about 1200 bps and only one byte at a time. Because there is no buffer in the Robonova and often considerable delay in execution in the robonova, plus the latency of single byte transactions on the PC, then communication performance is poor. Again this means that with standard RoboNova code and Robobasic, dynamic control of servos or sensors is not practical, since you will only get a few bytes transferred per second.
Still let us know how you decide. We can help with information on the Robonova side protocols and programming.
by Spiked3
Fri Feb 27, 2009 2:09 pm
The more I looked at it, the more I do not see it as a project 1 person would take on. The C#/.net aspect of it is ok with me, as I did that for several years as a Sr. Developer. Even the communication aspect seems workable to me. But the software on the RN-1 side would have to replaced, and I do not have enough embedded development experience to take on that part of it. I will take a longer look at some of the asm listings you have posted, to see if it's something I might be ok at doing.
In an ideal world, I see the RN-1 code would simply be a high performance 'moveTo' scheduler that dealt with synchronization of the servos, including gyro influence, and sending sensor data back to the PC. Buffering on the RN-1 side would be part of that code as well. Not so much for 'moves' as the RN-1 already does, but more for 'autonomous' behavior but with a bigger brain (and not really truly autonomous). But like I said, just wishful thinking.
In an ideal world, I see the RN-1 code would simply be a high performance 'moveTo' scheduler that dealt with synchronization of the servos, including gyro influence, and sending sensor data back to the PC. Buffering on the RN-1 side would be part of that code as well. Not so much for 'moves' as the RN-1 already does, but more for 'autonomous' behavior but with a bigger brain (and not really truly autonomous). But like I said, just wishful thinking.