In many respects, the difficult problem of synthetic aperture sonar is motion estimation of the platform since there exists no reasonably priced inertial measurement unit which can meet the location accurately requirements needed to generate high resolution imagery. Many beamforming codes estimate their motion using a displaced phase center technique. This technique is popular but makes some approximations that break down in long range systems and in significant motion environments. In this paper, we discuss the motion estimation algorithm used in ASASIN, a time domain back projection beamformer developed at Penn State Applied Research Laboratory. Our motion estimator assumes displaced phase centers but moves away from the phase center approximation. Rather than estimating the vehicle position delta ping-to-ping independently, we estimate the vehicle’s velocity and acceleration through all pings in 3 space. We use Google’s Ceres solver to solve the resulting non-linear equations and additionally add a regularization term to deal with missing and bad data samples. Finally, we show imagery formed using our algorithm.