Monitoring a system for anomalies is a common approach to detect conditions that may lead to failures and to take corrective actions. Such monitors must be carefully crafted by engineers with the domain knowledge to understand what could constitute abnormal behavior. This process becomes increasingly challenging as the system and its operating scenarios increase in complexity. It is unlikely that the system engineer will consider all possible scenarios. To alleviate this challenge, we propose an approach to automate the synthesis of monitors from the traces of robotic systems.