Drive-by/indirect monitoring of bridges demonstrated high practicality for their assessment and evaluation. In this study a drive-by monitoring methodology using a four-wheeled robot equipped with multi-modal sensors is developed. The method isolates the bridge vibration response from other factors like road roughness and vehicle dynamics by leveraging data from adjacent roadways. The methodology also explores the effect of varying the robot's trajectory on different sides over the bridge to determine optimal conditions for precise real-time mode identification. Experiments on a real-world bridge demonstrated the identification of up to four modes, with an average variation of 2% compared to reference monitoring data. Driving trajectories showed minimal impact on results, suggesting comprehensive identification.