Answered
My reinforcement learning simulation runs for only 0 steps and 0 times in Simulink. I am not getting any error messages so I cannot pinpoint the issue, so I decided to ask.
It likely has to do with the priority of execution of the data store blocks. I would look more into it, but honestly I think you...

11 months ago | 0

Answered
PPO and LSTM agent creation
Hi, With lstm policies, BOTH the actor and the critic should have lstm layers. That's why you are getting this error. LSTM po...

11 months ago | 0

Answered
Matlab reinforcement learning stop working after a while
125 episodes are not that many. Is it always freezing on that episode? Given that you are not getting any errors/crashes, my hun...

11 months ago | 0

Answered
DDPG with LSTM layer fails?
Hello, I see a couple of things wrong with the current architecture (could be more): 1) When you use the lstm layer, the input...

11 months ago | 0

| accepted

Answered
Could you help clarify the terminology and usage of Exploratory Policy and Exploratory Model in TD3 Reinforcement Learning
The answer above is correct with a small caveat: Even if UseExploratorypolicy is set = 0, the agent will still explore during t...

11 months ago | 0

| accepted

Answered
Load a pretrained neural network object in rlNeuralNetworkEnvironment
Hi Vasu, You can use a pretrained environment model with MBPO agent as follows: 1) Create a rlContinuousDeterministicTransitio...

11 months ago | 0

Answered
Initial condition setting mPC
This should be helpful: https://www.mathworks.com/help/mpc/ref/mpcstate.html

1 year ago | 0

Answered
is it possible to use neural network prediction model in Adaptive MPC ?
You should be able to do that, but we don't have an example to show the workflow unfortunately. You would first need to use aut...

1 year ago | 0

| accepted

Answered
Make Jacobian as a function in Simulink
I would first use automatic differentiation to get the A,B,C,D matrices using this command. Then you would need to take the gen...

1 year ago | 0

Answered
RL DDPG Actions have high oscillation
Hi, The noise options you are mentioning are only used during training and are essential for exploration. If the plots you are ...

1 year ago | 1

Answered
I want to know how to understand the sampling time ts of the agent in reinforcement learning, and how its value will affect the training and results?
Think of it as how often you need the agent to make decisions. Sometimes you need to make decisions very frequently (e.g. if you...

1 year ago | 0

| accepted

Answered
An Error in LSTM Network Training Using Reinforcement Learning Toolbox
I believe that both the actor and the critic need to be LSTM networks. In your case only the critic is

1 year ago | 0

| accepted

Answered
Linear Model Predictive Control with state feedback
The most straightforward way would be to assume that all your states are outputs, so set the C to be the identity matrix.

1 year ago | 0

Answered
How to soften hard constraints on states in non-linear MPC?
Here is an example that shows how to do that with multi-stage MPC: https://www.mathworks.com/help/mpc/ug/use-multistage-mpc-wit...

1 year ago | 0

Answered
RL: Continuous action space, but within a desired range（use PPO)
You can always clip the agent output on the environment side. PPO is stochastic so the upper and lower limits are not guaranteed...

1 year ago | 0

Answered
How to add custom environment for Reinforcement learning toolbox?
Why do you need a 3d world for this problem? Unless you consider the z dimension (e.g. if you do planning for UAVs), you only ne...

1 year ago | 0

Answered
Reinforcement Learning - PPO agent with hybrid action space
Hello, The workaround you suggested makes sense to me. Unfortunately though, bullet #3 is not currently supported. You cannot d...

1 year ago | 0

Answered
How to make the agent automatically pause after training to a certain number of turns
You can actually automate this process without having to stop training if you are using Simulink. There is an port 'Use external...

1 year ago | 0

Answered
Scaling in the QP formulation
I think it's likely a typo. Scaling factors should be included in Wu, Wdu and Wy along with the individual weights.

1 year ago | 0

| accepted

Answered
For a Non linear Model predictive control toolbox, how can I give it a mathematical model, which is a subsystem defined in Simulink?
Unfortunately, this is currently not supported but our development team is aware of this request. If you don't have an analytica...

1 year ago | 0

| accepted

Answered
Constraints violation in MPC design
The more constraints you are adding, the smaller the solutions space, the harder it gets for the solver to find a solution. If y...

1 year ago | 0

| accepted

Answered
Is there any way to graph the MPC cost function?
If you are designing the controller in MATLAB, all 'move' functions have an info argument which containts the cost value (see he...

1 year ago | 0

Answered
The code for nmpc is not working even if i put it in same folder ?
I wasn't able to reproduce the error. Both state and output functions must be in different .m files. I did that and put everythi...

1 year ago | 1

| accepted

Answered
Problem with LSTM and PPO reinforcement learning
Hi, When you set up LSTM networks, the input layer needs to be the sequenceInput layer. See for example here: https://www.math...

1 year ago | 0

Answered
QP formulation from the MPC toolbox
We are currently using the dense formula as you mentioned, but also working on adding support for sparse problems. The following...

1 year ago | 0

| accepted

Answered
Hierarchical Reinforcement Learning in Matlab
Hierarchical RL basically comes down to breaking a compelx problem down into smaller 'easier to train' problems, and potentially...

1 year ago | 0

Answered
Tune PI Controller Using Reinforcement Learning
I also replied to the other thread. The fullyConnectedPILayer is a custom layer provided in the example - you can open it and se...

1 year ago | 0

| accepted

Answered
I want to convert the PI controller in this instance to a PID controller, so can my network be changed to something like this?
That should work - you are essentially adding one more weight to the PI layer for the D term. Potential issues may arise when yo...

1 year ago | 1

| accepted

Answered
Deep reinforcement learning and TD3 algorithm in the PID control
Have you seen this example?

1 year ago | 0

| accepted

Answered
How do I stop the reinforcement learning training and ensure agent is saved when the Episode Manager Crashes. (Stop button on episode manager not accessible)
If you use the stop button, the agent you will get will have the parameters learned to that point. You can then continue trainin...

1 year ago | 0