This study provides a comparison of a linear and unscented Kalman filter to estimate the dynamics of a two-stroke opposed piston engine. These engines maintain several advantages over conventional internal combustion engines including reduced heat transfer, higher power to weight ratio, and a larger expansion ratio. Additionally, a crank angle phasing is introduced between the two cranks to improve scavenging efficiency. However, the coupling of the two crankshafts through a large geartrain can cause high amounts of noise and vibration harshness (NVH) and mechanical efficiency losses. By removing the geartrain and controlling the decoupled crankshafts with motor-generators, the NVH can be minimized while maintaining the benefits of crank angle phasing. To control the input torque of the motors to the engine, accurate estimations of the crankshaft position and dynamics are necessary. While the unscented Kalman filter exhibits lower estimation error, the filter is sensitive to model uncertainty relating to cylinder pressure, demanding further investigation of the robustness of the nonlinear filter.