Trajectory Prediction on an ego-bike with IMU and camera using a Multi-State Constraint Kalman Filter​

Trajectory prediction is a crucial task for ensuring the safety and reliability of transportation systems, including bicycles. While the Multi-State Constraint Kalman Filter (MSCKF) algorithm has been successfully applied to trajectory prediction for drones and cars, its effectiveness for predicting bicycle trajectories remains uncertain. In this study, we investigate the feasibility and accuracy of using the MSCKF algorithm for bicycle trajectory prediction. The aim is to utilize a cost-efficient setup, employing a simple low-cost camera and an inexpensive IMU. We compare the performance of the different algorithms on existing data sets to determine whether the MSCKF is a suitable algorithm. This study serves as a significant step to- ward the development of more effective and accurate trajectory prediction methods tailored specifically for bicycles.