HCESC Colloquium - Health Care Engineering Systems Center

HEALTH CARE ENGINEERING SYSTEMS CENTER
HCESC Colloquium Series
April 6, 2015
3:30 PM, CSL Auditorium-B02
Visual Navigation of Small-Sized Robots with Applications to Healthcare
Cang Ye, Ph.D.
Associate Professor
Department of Systems Engineering
University of Arkansas at Little Rock
http://ualr.edu/cxye/
The advancement of 3D Flash LIDAR Camera (FLC) technology is changing the way how
visual information is analyzed and utilized by a robot to understand and interact with its
environment. An FLC illuminates the scene in its field of view with a single laser pulse or
modulated infrared light and focuses the reflected light onto its focal plane array to
produce both intensity and range images of the scene. Being able to provide intensity and
range data with precise pixel-to-pixel match at a high frame rate, the FLC has potential in
drastically improving the autonomy of small robotic systems that have a myriad of
applications in healthcare.
The objective of my research is to develop FLC-based visual navigation algorithms for smallsized robots, including Simultaneous Localization and Mapping (SLAM) and object
recognition methods. I will first give an overview of my current research and then focus this
talk on the topic of SLAM. The proposed SLAM method comprises an egomotion estimation
algorithm, called Visual Range Odometry (VRO), and an Extended Kalman Filter (EKF). The
VRO simultaneously processes the FLC’s intensity and range data and estimates the
camera’s egomotion, which is then used as the motion model by the EKF to estimate the
camera’s pose in the navigation frame by tracking a set of visual features sampled from the
previous intensity images. Several new schemes are devised to improve the EKF’s
consistency and thus attain a better pose estimation accuracy than existing methods. In this
talk, I will discuss the applications of the proposed SLAM method in a co-robot cane for the
visually impaired and other healthcare robots. I will also discuss the extension of the visual
navigation methods to more affordable RGB-D cameras for healthcare applications.