An Integrated MMW Radar System for Outdoor Navigation

Abstract

This thesis research presents the design and evaluation of an integrated sensor system for obstacle detection and collision avoidance for automobile vehicles and mobile robot outdoor navigation. We performed evaluation tests for autonomous and manual driving modes. At the core of the system is a 77 GHz MMW radar sensor that is able to operate robustly even under adverse weather conditions. This sensor has a range of approximately 200 metres and uses a linear array of receivers and wavefront reconstruction techniques to compute range and bearing of objects within a horizontal field of view of 12 degrees. We discuss signal processing and calibration techniques and present results obtained from simulation and the real sensor. Clutter rejection and the ability to classify and differentiate between objects in different road lanes and off-road is improved by integration of the radar data with information from a road geometry perception system. Road geometry perception is based on either of two systems: the first one detects road lane markers or other features parallel to the road through a video camera and thereby computes vehicle position and orientation with respect to the current driving lane; the second system uses a digital road map in combination with a GPS and heading gyroscope unit to compute vehicle location and orientation on the road. Finally, we discuss the performance results obtained for both systems. The integrated sensor system can thus operate as a copilot to a human driver for speed control with respect to traffic density and as a safety monitor for lane changing. In addition, it is able to interact with other modules, such as road followers and higher level planners, for complete autonomous driving.