A University of Minnesota research team was recently awarded two grants totaling more than $3 million from the National Science Foundation's Cyber-Enabled Discovery and MRI Programs to create robotic devices and computer vision algorithms that will assist with the early diagnosis of children at risk of developing disorders such as autism, attention deficit disorder (ADD) and obsessive compulsive disorder (OCD).

The team, led by computer science and engineering professor Nikolaos Papanikolopoulos in the University's College of Science and Engineering, is developing robotic instruments that could observe and automatically analyze abnormalities in children's movements and behaviors. Researchers have been using the Xbox Kinect to track the subjects, but in the future the technology could be expanded. By using novel robots, such as robot pets and robotic sandboxes, equipped with specialized detectors and software, the researchers will analyze the probability of abnormalities based on facial expressions and body positions.

"Researchers and scientists believe that psychiatric disorders display subtle physical abnormalities in childhood well before the onset of a full disorder," said Papanikolopoulos. "We believe that we can use new computational tools, including computer vision and robotics, with a unique new computer vision algorithm to observe and detect abnormalities in motor and emotion in children to automatically analyze them for abnormalities."

Traditionally, experts have conducted psychiatric assessments using a visual rating system after watching videos of the subjects' motor movements and facial emotional expressions. Those expert ratings are subjective, and are limited to the observer's particular expertise. In addition, the method is costly.

This cross-disciplinary research seeks to create a diagnostic instrument for mental disorders that combines the fields of computer vision/robotics and computer science with child psychology and psychiatry.

Using these new tools, the research team members hope to be able to create more effective tools for detecting at-risk children at an earlier age. Ultimately, they hope to create a diagnostic framework, including workshops and tutorials, as well as demonstrations to distribute their work and make it more widely available.