top of page

Research Projects

Machine Vision Based Feature Detection for Apple Trees

In this project, I work on detecting different features of an apple tree. I am looking at the canopy area, canopy volume, canopy color, shoot lengths, trunk diameter, and yield for individual tree.

Team: Jostan Brown, Manoj Karkee, Joseph R. Davidson, Cindy Grimm, Ashley Thompson

Decision Support System for Nitrogen Need Assessment in Apple Trees

I am using different traits I find in the apple trees as features to identify the nitrogen needs on individual trees. I look at how these features relate to the nitrogen level in the canopy (with leaf nitrogen concentration as ground truth). I am also trying to integrate the vigor rating from experts in this field into the decision support system.

By implementing vision-based nitrogen need assessment technology, the possibility of precise management at the individual tree level is unlocked, leading to a novel opportunity to improve tree quality and assessment. This project has potential beyond nitrogen assessment, as it can be used in other crucial orchard management practices, including but not limited to, irrigation, spray applications, yield estimations, and more.

Team: Jostan Brown, Manoj Karkee, Joseph R. Davidson, Cindy Grimm, Ashley Thompson

Nitrogen in Trees

Autonomous Navigation for precision nitrogen spraying in orchard

I am using multiple sensor on a ground based platform : Warthog (Clearpath Robotics) as perception input to the robot and make it spatially aware. The robot takes in those input and is able to figure out the tree rows and its relative orientation in the row. The robot gets continuous datapoints and is able to navigate the orchard safely.

Team: Deven Biehler, Jostan Brown, Manoj Karkee, Joseph R. Davidson, Cindy Grimm, Ashley Thompson

autonomous orchard navigating robot

Shuttlecock Aerodynamics using Machine Vision

In this work, I was tracking shuttlecocks in a calibrated space using stereo vision on slow motion cameras. I calibrated a large volume and launched a shuttlecock using pnuematic cannon. The shuttlecock was tracked in both cameras to precisely track its 3D location. The trajectory of the shuttlecock was then used for analysing the aerodynamic properties of the shuttlecock. This method could be one of the alternatives for traditional wind turbine based analysis for characterization of aerodynamic properties.

shuttlecock tracking using automated machine vision
shuttlecock tracking

Team: Bin Lyu, Lloyd V. Smith

Team: Arghavan Talebanpour, Lloyd V. Smith

headform injuries using biclycle helmet

We looked at the linear and rotational accelerations using different headforms and with different helmets. We tried to quantify the differences in responses between different headforms during similar injuries.

bottom of page