The increased deployment of drone-based solutions, and their integration into various sectors, necessitate robust security measures. Machine learning-based intrusion detection systems (IDS) have emerged as a promising solution to safeguard the Internet of Drones (IoD). In this paper, our objective is to systematically map the research landscape of machine learning-based IDS for the IoD, and to categorize studies based on machine learning models, IDS types, datasets, software environments, and attack classifications. This study aims to identify trends, gaps, and opportunities in this emerging field. To this end, we conducted a systematic mapping study, reviewing 12 studies published between 2014 and 2024. We categorized and analyzed the selected studies to present a comprehensive overview of the current state of research. Our approach incorporated elements from systematic review guidelines, particularly defining the search strategy and conducting a quality assessment. The analysis revealed diverse machine learning models used in the IoD context. The studies predominantly explore network-based IDS as the primary type of intrusion detection system. The software environments were primarily based on Python and its libraries, while the datasets employed varied widely. In terms of attack classification, multi-class classification was the main focus. This diversity in methodologies and focus areas indicates a lack of standardization, highlighting the need for more consistent approaches in future research. The findings highlight the significant potential of machine learning-based IDS to improve IoD security by effectively identifying and mitigating cyber threats.