An autonomous localization and navigation system is designed for omnidirectional indoor mobile robots, which takes NVIDIA Jetson TX2 as upper computing platform and senses surrounding environment by lidar. Integrating wheel odometer and single axis inertial measurement unit, the designed system locates the robot by using selfadaptive Monte -Carlo algorithm basing on particle wave filter and implements autonomous navigation and obstacles avoidance by A-star algorithm and TEB algorithm. Results of Stage simulation and field experiment both show that the designed system can autonomously and accurately locate the robot, construct surrounding map with high accuracy, plan optimal path and help the robot avoid dynamic obstacles in time.