PPO agent applied to ACC model
显示 更早的评论
I'm applying the PPO algorithm to the ACC model in this DDPG example:Train DDPG Agent for Adaptive Cruise Control
How ever the agent went wrong at 50th episode with error:
An error occurred while simulating "rlACCMdl" with the agent "agent".
[varargout{1},varargout{2}] = simWithPolicy(this.Env,this.Agent,simOpts);
[varargout{1:nargout}] = runImpl(this);
[varargout{1:nargout}] = run(task);
[this.Outputs{1:getNumOutputs(this)}] = internal_run(this);
runDirect(this);
runScalarTask(task);
run(seriestaskspec);
run(trainer);
train(this);
TrainingStatistics = run(trainMgr);
Caused by:
Invalid input argument type or size such as observation, reward, isdone or loggedSignals.
Standard deviation must be nonnegative. Ensure your representation always outputs nonnegative values for outputs that correspond to the standard deviation.
Is there anybody could solve this problem? Thanks for your help!
1 个评论
qun wang
2021-10-26
您好，请问这个问题你解决了嘛？我最近也在用PPO遇到类似的问题，可否讨论一下？期待回复
回答（1 个）
Emmanouil Tzorakoleftherakis
2020-9-3
0 个投票
Hello,
Can you make sure that you set up your actor following a structure similar to this one? It seems that your variance path is not set up properly and gives negative values.
类别
在 帮助中心 和 File Exchange 中查找有关 Reinforcement Learning 的更多信息
2020-8-22
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!
Translated by 
