Real-Time Monocular Visual Odometry for On-Road Vehicles with 1-Point RANSAC

The first biggest problem in visual motion estimation is data association; matched points contain many outliers that must be detected and removed for the motion to be accurately estimated. In the last few years, a very established method for removing outliers has been the "5-point RANSAC" algorithm which needs a minimum of 5 point correspondences to estimate the model hypotheses. Because of this, however, it can require up to thousand iterations to find a set of points free of outliers. In this talk, I will show that by exploiting the non-holonomic constraints of wheeled vehicles (e.g. cars, bikes, mobile robots) it is possible to use a restrictive motion model which allows us to parameterize the motion with only 1 point correspondence. Using a single feature correspondence for motion estimation is the lowest model parameterization possible and results in the most efficient algorithm for removing outliers: 1-point RANSAC. The second problem in monocular visual odometry is the estimation of the absolute scale. I will show that vehicle non-holonomic constraints make it also possible to estimate the absolute scale completely automatically whenever the vehicle turns. In this talk, I will give a mathematical derivation and provide experimental results on both simulated and real data over a large image dataset collected during a 25 Km path.