Aspect Sentiment Triplet Extraction (ASTE) extends Aspect-Based Sentiment Analysis (ABSA) by identifying triplets consisting of aspect terms, opinion expressions, and sentiment polarities. The original ASTE-RL study demonstrated that incorporating Reinforcement Learning (RL) into a BERT-based model improved F1 performance by approximately 1 % on English datasets compared to using a BERT-only model. Motivated by this finding, our study investigates whether a similar improvement can be achieved in the context of Bahasa Indonesia. In addition, no prior work has explored RL-based ASTE for Bahasa Indonesia. We adapt the ASTE-RL framework by integrating IndoBERT embeddings within a hierarchical RL model that combines sentence-level sentiment classification with aspect-opinion extraction. Experiments on an Indonesian benchmark dataset show that the proposed model achieves an F1-score of 80.33 % on the development set and 72.63 % on the test set. Although the gap between development and test performance suggests potential overfitting, our findings provide the first empirical evidence of applying RL-based ASTE to Bahasa Indonesia. © 2025 IEEE.