7/17/2015 David Robertson, CSL 3 min read
Written by David Robertson, CSL
Along with doing the laundry and taking out the trash, mowing the lawn is an inescapable chore of everyday life (at least for anyone with a yard in the summer).
Sweltering heat, pesky bugs, and lack of time might deter someone from venturing outside to cut the grass. But technology that will make it possible to circumvent the whole job is nearly finished at the Illinois' Coordinated Science Lab.
Junho Yang, a PhD candidate in mechanical science and engineering, is working with Professor Seth Andrew Hutchinson, Aerospace Assistant Professor Soon-Jo Chung,and agriculture implement manufacturer John Deere, on “an omnidirectional-vision-based system to detect the containment status of a robotic lawn mower.” John Deere is funding the project.
For Yang, this project relates closely to his previous work for the Office of Naval Research. He contributed to estimating the location of a Unmanned Aerial Vehicle (UAV) that operates in a riverine environment and produces a 3-D point-feature-based map of the area mostly with a monocular camera.
While robotic lawn mowers are currently available in the marketplace, existing models require the setup of boundary wires so that the mower knows where to operate. For someone with an expansive yard, driving stakes into the soil and running wires is still an inconvenience.
With the new omnidirectional-vision-based system, there would be fast and easy set up. Some of the biggest challenges to overcome are ensuring the mower makes clean, straight paths and monitoring disruptions, such as a rock or small object that could appear in the line of travel.
“Our system has the potential to enable no-infrastructure installations and perform positioning of the robot for non-random path mowing,” Yang said. “We are trying to make the vision processing and estimation more robust to disturbance and noise.”
Currently the prototypes are made from a John Deere Tango lawn mower, an omnidirectional camera, and an inertial measurement unit. They are being tested at the John Deere Technology Innovation Center at the Research Park at the University of Illinois. John Deere employees are also testing them in their front yards, Yang said.
After three years of hard work,Yang and his colleagues presented the project at the IEEE International Conference for Robotics and Automation in May, to positive reception.
The team is working to have the project finished this summer. Robotic lawn mowers equipped with a vision system should be available to the masses in the future, but details about that are confidential at present.
“Our vision-based estimation technology will eventually help autonomous robots to help us in our daily life,” Yang said.