|
Computer vision is a burgeoning field of research in South Africa and deals with the science and technology that allows machines to see.
It is an essential component of field robotics that involves extracting information from images and interpreting it for use by robots.
Natasha Govender says this new research field is exciting and very challenging.
She is a senior researcher with the mobile intelligent autonomous systems (MIAS) group that resides within the CSIR modelling and digital science research domain. MIAS focuses on field robotics.
She says although the computer vision field is relatively new in South Africa, there is a lot of research being conducted at present.
Research
Govender’s main area of research in the computer vision field is object recognition and object pose estimation. 3D object recognition refers to the ability to recognise a specific 3D object in a cluttered environment, while object pose estimation determines the position and orientation of the object relative to a specific co-ordinate system.
In her current project on a Barrett arm manipulator, Govender explains that a camera has been mounted on the arm of a robot to capture images of the environment. “These images are then used to extract information about the objects and the environment. This information allows the arm to manipulate objects in the environment and is also used for collision detection,” she says.
A robotic arm of this nature could also be used to help perform certain tasks that are deemed extremely dangerous for humans to carry out, such as in areas where there is nuclear waste or chemicals.
An active vision component will also be added to the system. She explains: “An active vision system is one that can manipulate the viewpoint of the camera or of an object in order to obtain more useful information,” adding that this reduces time and improves accuracy for object recognition systems. These benefits will have a huge impact on complex systems, which require these tasks to be performed quickly with a fair degree of accuracy.
Her work forms part of her PhD research on incorporating active vision into object recognition and pose estimation. She is currently in the second year of her PhD for which she is registered at Oxford Brookes University in the United Kingdom.
Govender, a former CSIR bursar, is passionate about human capital development and spends a lot of her spare time mentoring students and doing volunteer work. She is also a member of the CSIR squash and volleyball clubs.
She has recently been selected as one of only 120 PhD students worldwide to attend an international computer vision summer school in Sicily, Italy in July this year.
Govender has a number of journal publications, international peer-reviewed articles and local publications to her name.
CSIR Strategic Communications and Stakeholder Relations :
Mzimasi Gcukumana, email: MGcukumana@csir.co.za
|
.jpg)
