Sensor Fusion for Octagon – an Indoor and Outdoor Autonomous Mobile Robot

Abstract

Robot and self-driving vehicle navigation and localization problems have been a hot topic in recent years. With the significant development of ranging sensors such as three-dimensional Light Detection and Ranging (LiDAR), Radio Detection and Ranging (RADAR), depth cameras (Microsoft Kinect and Intel RealSense), robot and autonomous vehicles have improved their understanding of the surrounding environment. However, a single sensor can only provide limited information, and robot multi-sensor fusion has offered new perspectives for navigation and localization. Sensor fusion technology can solve two complex problems in robot navigation. Typically, mobile robots only detect above-ground objects for obstacle avoidance. But in the real world, a hole or gap along the robot path can cause damage to the robot and needs to be avoided. Typically, robot navigation uses a single approach for the solution of the localization problem by using an indoor map or GPS. However, in some cases, robots have to navigate between indoor and outdoor environments and must use different solutions for reliable operation. This paper presents an industry-level autonomous mobile robot that uses the sensor fusion method to solve indoor and outdoor navigation tasks. For safe navigation, a below-ground obstacle detection algorithm is presented that is based on point cloud data provided by the Kinect depth camera, which can then be added to the location of above-ground obstacles provided by the 2D LiDAR SLAM algorithm. A navigation source switching algorithm is designed for switching robot navigation systems using an indoor map and outdoor GPS.