In this paper, we present a tomographic reconstruction method to reduce a tomographic error, multi time-stepreconstruction, for a wide-eld adaptive optics (WFAO). Based on the frozen ow assumption, we can computethe time evolution of measurements from wave-front sensors (WFS) at previous time-steps with using windinformation. Our idea is to reduce the tomographic error by using the measurements at both the current andprevious time-steps simultaneously. We also develop a method to estimate wind speed and direction at eachaltitude from temporal correlations of phase distortion pattern reconstructed by a classical tomography. Weevaluate the performance of the method by a laboratory experiment with the RAVEN, a multi-object adaptiveoptics (MOAO) technical and science demonstrator. In the laboratory experiment, our wind estimation methodcan estimate wind speeds and directions of multiple layers. By the multi time-step reconstruction method, theensquared energy in a 140 mas box increases about 3{5% compared with a classical tomographic reconstruction.

Raven is a Multi-Object Adaptive Optics science demonstrator which has been used on-sky at Subaru telescope from May 2014 to July 2015. Raven has been developed at the University of Victoria AO Lab, in partnership with NRC, NAOJ and Tohoku University. Raven includes three open loop WFSs, a central laser guide star WFS, and two science pick-off arms feeding light to the Subaru IRCS spectrograph. Raven supports different AO modes: SCAO, open-loop GLAO and MOAO. This paper gives an overview of the instrument design, compares the on-sky performance of the different AO modes and presents some of the science results achieved with MOAO.