Obstacle Avoidance using mmWave Radar Imaging System

Abstract

This paper considers a path generating for obstacle avoidance based on millimeter-Wave (mmWave) radar imaging system with a rapidly exploring random tree (RRT) algorithm. The position of obstacle can be detected by the IWR1443 radar imaging sensor. The shape of the obstacle is assumed that is a cylindrical shape. The data of the sensor is represented in the form of a position, which has no information about shape. The proposed algorithm has two stages, which are separation and estimation. The obstacles are separated by K-mean with the K-selection algorithm. The position of obstacles is estimated by using a circle-fitting method in the XY-coordinate. Since the estimated radius has an error, we propose multiplying and summation parameters to compensate for the estimated radius. By adjusting the radius, we could design the possible path based on RRT*. Our results show the path between the starting point to the goal point with two obstacles. The path of the RRT* can avoid the obstacle with the information from the IWR1443 sensor. This algorithm is used to avoid obstacles in 2D space. The compensate parameter is defined by the environment, the robot, and the user.