onder

45 Reputation

7 Badges

8 years, 169 days

MaplePrimes Activity


These are replies submitted by onder

@Joe Riel Hi Joe, When I cilck MapleSim Examples and open 5 DOF robot I am able to see the offsets and the xyz directions of the rigid body frames in the Base but when I click the rigid body frames of the linkages, the inspector pane is empty. I cannot see the offsets and xyz directions. I was using MapleSim 2014 and I did not have such a problem, now I use MapleSim 2015, I have this problem.

@Thomas Richard Hi Thomas, thank you for your answer. You understood the question correctly. I can see the 5 DOF robot Base Rigid Body Frame offsets (both the lengths and directions) but I cannot see the linkages' offsets and xyz directions. The inspector pane is empty for the rigid body frames. 

@Joe Riel 

Thanks for the information. Then, in this case how can I only extract the real part of the complex signal from this model (above)?

 

Hi,

I am seeking answer the question that you asked about MapleSim plot font size. Have you got anything about that? If you have, could you share it with me please?

Best

Onder

Hi Graham,

Thanks for your reply. If the system has just one torgue input for the double inverted pendulum, then it linearizes the system. However, I am unable to design a PID controller because a message appears saying that the system must be single input single output. Just I want to linearize the double inverted pendulum and design a PID controller for the both joints. Here is the model DoublePendulumPIDTun.msim. that I consider.

Thanks for your help...

Hi Graham,

Thanks for your reply. If the system has just one torgue input for the double inverted pendulum, then it linearizes the system. However, I am unable to design a PID controller because a message appears saying that the system must be single input single output. Just I want to linearize the double inverted pendulum and design a PID controller for the both joints. Here is the model DoublePendulumPIDTun.msim. that I consider.

Thanks for your help...

Hi Graham,

I have an artificial intelligence algorithm which is applied to a simple two legged robot. Something strange is happening. The reward function is (1-cos(swingAngle))/0.1077 where swing angle converges to -0.38. In this case I was expecting to see the reward value at steady state 0.6624. However, the reward is 0.995. I changed the reward function but the result is still miscalculated. (Please see the file. instantRewardSwingLeg is the reward function modelica custom component, reward plot is the reward figure.). Could you advise me how to get rid of this problem?

Another thing is I need to reinitialize the both legs' angles/velocities when the swing leg hits the ground and the angles reach the value  (both swing and support) that are appropriate for walking. Basically based on the number of measurements I need to reinitialize the both legs for the next steps. Could you help me for this please.

Here is the file RLHumanoidRobotFull3.msim

Thanks for your help...

Hi Graham,

I have an artificial intelligence algorithm which is applied to a simple two legged robot. Something strange is happening. The reward function is (1-cos(swingAngle))/0.1077 where swing angle converges to -0.38. In this case I was expecting to see the reward value at steady state 0.6624. However, the reward is 0.995. I changed the reward function but the result is still miscalculated. (Please see the file. instantRewardSwingLeg is the reward function modelica custom component, reward plot is the reward figure.). Could you advise me how to get rid of this problem?

Another thing is I need to reinitialize the both legs' angles/velocities when the swing leg hits the ground and the angles reach the value  (both swing and support) that are appropriate for walking. Basically based on the number of measurements I need to reinitialize the both legs for the next steps. Could you help me for this please.

Here is the file RLHumanoidRobotFull3.msim

Thanks for your help...

@Joe Riel Hi Joe, I managed to eliminate the problem that I specified above for you. We are publishing the preliminary results that we obtained so far in a conefernence called '2012 International Conference on Advanced Mechatronic Systems, Japan'. I would like you to be the co-aouther of the paper, if it is OK for you. I am planning to write a small section about MapleSim/Modelica. If you just read that section and give me feedback, that would be great. After this publication, we are plaaning to continue fruther applications...

One more question, I need to get the matrices at the last time point of the simulation from the output of criticParamUpdate and actorParamUpdate Modelica blocks. How can I obtain these matrices which are the determined at the last time point? Then, I will draw 3D plots. If I get these matrices, I think it could be easier for me to plot 3D plots in Matlab...

Thanks for your help...

@Joe Riel Hi Joe, I managed to eliminate the problem that I specified above for you. We are publishing the preliminary results that we obtained so far in a conefernence called '2012 International Conference on Advanced Mechatronic Systems, Japan'. I would like you to be the co-aouther of the paper, if it is OK for you. I am planning to write a small section about MapleSim/Modelica. If you just read that section and give me feedback, that would be great. After this publication, we are plaaning to continue fruther applications...

One more question, I need to get the matrices at the last time point of the simulation from the output of criticParamUpdate and actorParamUpdate Modelica blocks. How can I obtain these matrices which are the determined at the last time point? Then, I will draw 3D plots. If I get these matrices, I think it could be easier for me to plot 3D plots in Matlab...

Thanks for your help...

@Joe Riel We added noise in order to find the optimal policy. As optimal learning occurs, the amount of the noise reduces. Probe 1 measures the position of the pendulum. If you run the simulation for a couple of times, then you will see position state (probe 1) converges to stable point, std and n go to zero as well. (I got these results for several times) Each run should give the same simulation results, but even though I assigned all the initial values, sometimes it gives different results. (Unstable results as in your case). What could be the reasons for these changes? Is it because of simulation setting 'adaptive=false'" ? As lang as the generated states by the pendulum same and the assigned initial values are same, then the artificial intelligence algorithm must generate same results. It seems to me something is randomly affecting the generated states by the pendulum which causes the divergence in the artificial algorithm outputs. (We might publish these results in a robotic/artificial intelligence conference, but we are not certain abot the results) This is working perfectly in Matlab. Thanks for your help...

Follow up;

For example for the first run in MapleSim5,  I got this result which is unstable

And for another run I got this onewhich is table. I am expecting to see the same results at every run as long as I use the same initial states and noise. It seems pendulum is generating slightly different states yielding difference in results...

@Joe Riel We added noise in order to find the optimal policy. As optimal learning occurs, the amount of the noise reduces. Probe 1 measures the position of the pendulum. If you run the simulation for a couple of times, then you will see position state (probe 1) converges to stable point, std and n go to zero as well. (I got these results for several times) Each run should give the same simulation results, but even though I assigned all the initial values, sometimes it gives different results. (Unstable results as in your case). What could be the reasons for these changes? Is it because of simulation setting 'adaptive=false'" ? As lang as the generated states by the pendulum same and the assigned initial values are same, then the artificial intelligence algorithm must generate same results. It seems to me something is randomly affecting the generated states by the pendulum which causes the divergence in the artificial algorithm outputs. (We might publish these results in a robotic/artificial intelligence conference, but we are not certain abot the results) This is working perfectly in Matlab. Thanks for your help...

Follow up;

For example for the first run in MapleSim5,  I got this result which is unstable

And for another run I got this onewhich is table. I am expecting to see the same results at every run as long as I use the same initial states and noise. It seems pendulum is generating slightly different states yielding difference in results...

@Joe Riel I thought they do not cause any problem since they are not part of the simulation. I deleted them. The problem is now whenever I re-run the simulation I have different simulation results. I assigned all the important initial states for the modelica blocks and this reduced the amount of changes in results between each simulation. However, the differences in each simulation in terms of results is still significant. For example, for a simulation, the results are stable and position state converges to 0.1. In another run it converges -0.12 or it is unstable. I thought this is due to simulation setting 'adaptive=true', hence changed it to' false'. A message 'Maximum number of events (100) has been reached at time 1.2' appeared. Could you advise me on possible reasons for having different simulation results and possible solutions please. The file test1.msim. Thanks for your help...

@Joe Riel I thought they do not cause any problem since they are not part of the simulation. I deleted them. The problem is now whenever I re-run the simulation I have different simulation results. I assigned all the important initial states for the modelica blocks and this reduced the amount of changes in results between each simulation. However, the differences in each simulation in terms of results is still significant. For example, for a simulation, the results are stable and position state converges to 0.1. In another run it converges -0.12 or it is unstable. I thought this is due to simulation setting 'adaptive=true', hence changed it to' false'. A message 'Maximum number of events (100) has been reached at time 1.2' appeared. Could you advise me on possible reasons for having different simulation results and possible solutions please. The file test1.msim. Thanks for your help...

@Joe Riel Sorry Joe, I attached the wrong file. The problem that I currently face is when I re-run the simulation, I have different simulation results. I assigned all the important initial values for the modelica blocks and this reduced the amount of the changes in the results for each simulation. However, sometimes differences between each simulation is still significant. I suppose this is due to simulation settings 'adaptive=true'. I changed it to false and assigned 0.01 time step, then it gives a warning which tells that the maximum number of events (100) has been reached and gives the results up to 1.2s. Please advise me on this issue. Here is the file test1.msim. Thanks for your help...

1 2 3 Page 1 of 3