Abstract: An active data collection mode control system to reduce crowd sourcing signal data collection required for fingerprint database (FPDB) maintenance is provided. The system includes a mobile device that supports a survey mode, a localization mode, and a crowd-sourcing mode, and a server that receives data from the mobile device, generates and updates an FPDB, and controls a data collection mode.

Abstract: Provided is an atypical environment-based location recognition apparatus. The apparatus includes a sensing information acquisition unit configured to, from sensing data collected by sensor modules, detect object location information and semantic label information of a video image and detect an event in the video image; a walk navigation information provision unit configured to acquire user movement information; a metric map generation module configured to generate a video odometric map using sensing data collected through a sensing information acquisition unit and reflect the semantic label information; and a topology map generation module configured to generate a topology node using sensing data acquired through the sensing information acquisition unit and update the topology node through the collected user movement information.

Abstract: Provided is an atypical environment-based location recognition apparatus. The apparatus includes a sensing information acquisition unit configured to, from sensing data collected by sensor modules, detect object location information and semantic label information of a video image and detect an event in the video image; a walk navigation information provision unit configured to acquire user movement information; a metric map generation module configured to generate a video odometric map using sensing data collected through a sensing information acquisition unit and reflect the semantic label information; and a topology map generation module configured to generate a topology node using sensing data acquired through the sensing information acquisition unit and update the topology node through the collected user movement information.

Abstract: Provided is a segmentation and tracking system based on self-learning using video patterns in video.

Abstract: Provided are a system and method for active data collection mode control for reducing crowd-sourcing signal data collection required for fingerprint database (FPDB) maintenance. The system for active data collection mode control for reducing crowd-sourcing signal data collection required for FPDB maintenance includes a mobile device configured to support a survey mode, a localization mode, and a crowd-sourcing mode and a server configured to receive data from the mobile device, generate and update an FPDB, and control a data collection mode.

Abstract: An apparatus and method for controlling virtual training simulation are disclosed herein. The apparatus for controlling virtual training simulation includes a posture recognition unit, a posture information convergence unit, a location recognition processing unit, and a movement device control unit. The posture recognition unit recognizes the posture of a trainee based on the image information of the inside of a virtual training field and motion sensor information. The posture information convergence unit generates converged information by converging the results of the recognition of the posture of the trainee. The location recognition processing unit estimates the current location of the trainee based on the converged information. The movement device control unit controls an omnidirectional movement device in which the trainee is placed using information about the control state of the omnidirectional movement device and the current location of the trainee.

Abstract: Disclosed are a method and an apparatus for predicting human motion in a virtual environment. The apparatus includes a motion tracking module configured to estimate a human pose of a current time step based on at least one piece of sensor data and a pre-learned motion model, and a motion model module configured to predict a set of probable human poses in the next time step based on the motion model, the estimated human pose of the current time step, and virtual environment context information of the next time step. A sense of immersion of the virtual environment may be maximized.