If you used the same tuning parameters, then yes. train is designed to let you use the same code to fit different models and also to tune those models.
In this case, depending on how you call it, train might not evaluate the value of mtry that your call to randomForest used. If that's the case, then the results would be different.
It might help if you read the paper on caret. It's a bit old but will explain what the package is meant to do. There is also a ton of documentation that might help.