Astronauts exercise during long space flights to stay fit and prevent adverse health problems bone density loss. Monitoring their exercise sessions by camera and doing kinematic analysis will provide valuable insights. However, current bulky multi-camera systems are unsuited to spacecraft requirements, and single video camera systems present challenges of lack of depth information, and partial occlusion of parts of the body. To address this, IAI and its collaborator, C-Motion, have been awarded a follow-on contract entitled “ESPRIT: Exercise Sensing and Pose Recovery Inference Tool.” A preliminary design of ESPRIT, which is a stereo camera system with a small footprint in terms of size, weight, power consumption and setup time, has been completed. ESPRIT detects markers placed on the body or clothing and other image features and recovers 3D kinematic information of the body pose. ESPRIT innovatively handles partial occlusion issues by relying on multiple feature cues, strong prior knowledge and modeling of the human body, pose, dynamics, and appearance, and on advanced machine learning and statistical inference techniques to achieve robust and accurate pose estimation. ESPRIT uses the statistical sampling-based Markov chain Monte Carlo method to compute a global optimization of the human pose trajectory. Motion capture of several exercises, including walking, curling and dead lifting, have already been demonstrated. Algorithms for marker detection and tracking, feature extraction, stereo matching of features, marker labeling and pose estimation have also been developed. Future work will focus on enhancement of the algorithms, development of an ESPRIT prototype, detailed performance evaluation, and delivery of the prototype for testing and demonstration.