Human Pose Estimation for Care Robots Using Deep Learning

Human Pose Estimation for Care Robots Using Deep Learning


Jul 12, 2017 @ 08:12 |


A research group led by Professor  at Toyohashi University of Technology, has developed a method to estimate various poses using deep learning with depth data alone.


Although it requires a large volume of data, the group has realized a technology which efficiently generates data using computer graphics and motion capture technologies. This data is freely available, and expected to contribute to the progress of research across a wide range of related fields.

Expectations for care robots are growing against the backdrop of declining birthrates, an aging population, and a lack of care staff. As an example, for care at nursing homes and other such facilities, it is anticipated that robots will check the condition of the residents while patrolling the facility. When evaluating a person’s condition, while an initial estimation of the pose (standing, sitting, fallen, etc.) is useful, most methods to date have utilized images. These methods face challenges such as privacy issues, and difficulties concerning application within darkly lit spaces. As such, the research group (Kaichiro Nishi, a 2016 master’s program graduate, and Professor Miura) has developed a method of pose recognition using depth data alone .

human poses
 Procedure of generating learning data

For poses such as upright positions and sitting positions, where body parts are able to be recognized relatively easily, methods and instruments which can estimate poses with high precision are available. In the case of care, however, it is necessary to recognize various poses, such as a recumbent position (the state of lying down) and a crouching position, which has posed a challenge up until now. Along with the recent progress of deep learning (a technique using a multistage neural network), the development of a method to estimate complex poses using images is advancing. Although deep learning requires preparation of a large amount of training data, in the case of image data, it isrelatively easy for a person to see each part in an image and identify it, with some datasets also having been made open to the public. In the case of depth data, however, it is difficult to see the boundaries of parts, making it difficult to generate training data.

human poses
Example of generating training data for various sitting positions. First row: body part label images, Second row: depth data

As such, this research has established a method to generate a large amount of training data by combining computer graphics (CG) technology and motion capture technology . This method first creates CG data of various body shapes. Next, it adds to the data information of each part (11 parts including a head part, a torso part, and a right upper arm part), and skeleton information including each joint position. This makes it possible to make CG models take arbitrary poses simply by giving the joint angles using a motion capture system. 


  • Keywords:  Human Pose Estimation, Robots, Deep Learning, computer graphics,  Technology.
  • Source: Toyohashi University of Technology
  • Image: Toyohashi University of Technology

 

Leave a Reply

Your email address will not be published. Required fields are marked *