Visual Odometry Research
Visual Odometry Research
In Professor Aaron Johnson's Robomechanics Lab, I conducted research in visual odometry for legged robots whose movement consists of pitch motion, especially when bounding. I explored the effect of pitch motion on visual odometry estimation when the robot was moving straight forward. Visual odometry confused pitch motion for some presence of vertical displacement which was expected. I also explored the effect of pitch frequency on the accuracy of visual odometry calculation. I found that there was a max frequency for each value of pitch angle at which the x-direction estimation began to lose track. The max pitch frequency decreased logarithmically with increasing pitch angle. I presented all of my findings during the Meeting of the Minds on Wednesday, May 9, 2018. You can access the executive summary and the presentation slides.