The cleaning robot can perform cleaning tasks autonomously and intelligently, thus saving people from boring and heavy physical labour. Autonomous positioning is one of the indicators for intelligent level of cleaning robots. In this paper, dead-reckoning method based on encoder is adopted. The kinematics model of the robot is established to optimize the existing cleaning robot positioning method by using arc model. The positioning effect of the cleaning robot is tested by experiment. Through experiments, it is found that there are two main problems in the estimation of the track. One is that during the turning process, the calculation of the angle is affected by the slip, which causes a large deviation. Another is that the displacement deviation of linear motion gradually increases with the increase of the driving distance due to system error. Based on the above analysis, a positioning scheme is proposed. Gyro sensor and RFID tag are introduced to correct the angular deviation and assist positioning and improve the positioning accuracy of the cleaning robot.

Autonomous quadruped robots require localization and mapping for navigation. Different from mobile robots, quadruped robots need local dense maps with more detailed information for motion planning. A key challenge, limited by the capacity of computers on robots, the perception systems have to balance the properties of the speed and accuracy to ensure that robots reach their destinations quickly and safely. In this paper, we propose a complete solution for the autonomous movement of quadruped robots. Localization is achieved by the Extended Kalman Filter which is updated by the results from laser based the ICP algorithm. Dense maps are created by raw ranger sensor data and accumulated base on accurate localization results. And then, dense maps are transformed to the occupancy grid maps which contain free and occupied space information for planning. For implementations, we demonstrate the effectiveness of our approach in a complex outdoor environment.

No abstract