A three-wheeled omnidirectional SLAM robot is designed and developed independently. The core of the robot is a NVIDIA Jetson Nano with a Robot Operating System (ROS) wrapper—
Melodic—over the Ubuntu 18.04 OS. The SLAM algorithms are implemented in the Jetson Nano including mapping and localization. Additionally, a Raspberry Pi 3 B+ is used to implement low level algorithms including encoder signal processing, motion sensor signal processing, PID control of motors and robot. The Jetson Nano communicates with the Raspberry Pi under the ROS protocol, and remote connections can be built between a host computer and the two development boards respectively through WiFi.
The robot can run in holonomic mode and nonholonomic mode. And by using an electronic compass integrated into the motion sensor, the robot can also run in compass mode to face north. The simulation below uses a two-wheeled robot to build a map using the slam_gmapping package, and the designed robot is tested to build a map in the real world environment.
Comments