Complete Odometry Estimation of a Vehicle Using Single Automotive Radar and a Gyroscope

In this paper, we propose an algorithm for complete odometry of a vehicle on a horizontal plane., that is., estimation of linear velocity vector (forward and sideslip speeds) and angular speed of a vehicle. The vehicle is equipped with an automotive Radar sensor and a vertical gyro. The Radar sensor provides radial speed and azimuth angle of number of objects in the environment. We first derive the kinematic constraints imposed on the vehicle motion and stationary points in the environment. Using the constraints we classify the points detected by the Radar to stationary and non-stationary points. It is known that using data from a single Radar., the abovementioned constraints are singular. Previous works have thus proposed the use of more than one Radar sensor., or they have neglected the sideslip speed. In our work, we then use the Radar data of the stationary objects and a gyro data to solve an optimization algorithm to calculate vehicle odometry. Experimentation has been performed with a non-road vehicle driven on a straight path and on a circular path. We report our findings and show efficacy of the algorithm in comparison to the state of art [8] as well as wheel odometry and a complete navigation solution (including GNSS) as the reference path.