Camera Calibration: Helping Robots See Straight
December 19, 2020 at 6:00:00 AM
Team 971 (Mountain View High School)
Dr. Jim Ostrowski is the Chief Technology Officer for Blue River Technology, a Sunnyvale-based company working to bring tools from computer vision and robotics to bear in agriculture. Blue River Technology, which was acquired by John Deere in September, 2017, is focused on developing See & Spray technology that provides real-time weed detection and spraying and can reduce herbicide usage by 90%. Before that, Jim worked for Evolution Robotics, which later became Evolution Robotics Retail, before being acquired by Datalogic. Jim's field of expertise includes mobile robotics, nonlinear dynamics, and vision-based recognition and control using classical tools and leveraging Deep Learning. Before joining the start-up world, Jim was at the University of Pennsylvania, where he was an Associate Professor in the Department of Mechanical Engineering and a member of the General Robotics, Sensing, Automation and Perception (GRASP) Laboratory. He holds a Ph.D. in Mechanical Engineering from the California Institute of Technology. He is a former Associate Editor for the journal IEEE Transactions on Robotics and Automation, was the recipient of the NSF CAREER Young Investigator Award for his work in robotics, and received a Datalogic VIP award for his technical work on the LaneHawk product.
About the Presentation
In this talk, I will provide an overview of intrinsic camera calibration and give details on how 971 has approached this problem. This will start with an introduction to how objects are projected into camera images, and walk through the use of OpenCV tools and Charuco target patterns to do camera calibration. The goal is to make sure that robots are seeing the world correctly-- straight lines stay straight, and we can accurately measure where objects are in the world.