It make sense to repeatedly evaluate how an algorithm works for a particular scenario or environment. rlR provides the function rlR::repExperiment to serve this need. One could also use different cores to execute seperate experiment.
library(doMC)
## Loading required package: foreach
## Loading required package: iterators
## Loading required package: parallel
registerDoMC(5)
# or
library(doParallel)
cl <- makeCluster(5)
registerDoParallel(cl)
res = repExperiment(sname = "CartPole-v0", aname = "AgentDQN", conf = getDefaultConf("AgentDQN"), nrep = 5, nepi = 5)