1 Introduction
Map creation is a fundamental and very important technology for mobile robots. It is the basis for technologies such as navigation, localization and path planning of mobile robots in unknown environments [1]. We can understand the problem of map creation for mobile robots as follows: Starting from any time and any position, a mobile robot should be able to explore the environment through its sensors, obtain relevant environmental information, and thus build a suitable map [2].
There are two common map representation methods for mobile robots. One is the occupancy grid method, which uses the probability value of each grid being occupied to represent environmental information, but the accuracy of grid maps is not high [3]. The other is the geometry-based representation method, which uses line segments and arcs to represent environmental information. Geometric maps have a small storage capacity and can be directly applied to the navigation and path planning of mobile robots [4]. Currently, ultrasonic and infrared sensors are commonly used in mobile robot navigation research at home and abroad, but they generally have the problem of low accuracy. Compared with them, laser sensors have the advantages of fast data acquisition speed and high accuracy [5].
This paper uses laser sensors to collect specific planar environmental information of the indoor environment. The data points in this information are piecewise fitted into straight lines using the least squares method. The fitted lines are then filtered by calculating the distance variance to obtain effective line segments. These effective line segments are merged and represented using geometric information to obtain the current local map. Finally, the local maps obtained from each frame are matched to obtain the global map.
2. Creation of Unknown Environment Maps
There are two types of environment map creation: local environment map creation and global environment map creation.
2.1 Local Map Creation
Figure 2. Relationship between local and global coordinate systems
2.3 Global Map Creation
The first frame of the local map is selected as the current global map. After extracting the straight line features of the current frame (i.e., creating the local map), the obtained local map is matched with the current global map to obtain the relationship between the line segments, thereby obtaining the relevant feature points between them. Then, the offset matrix between the local map and the global map is calculated using the feature points, and the local map is merged into the global map to update the global map.
3. Experimental Results and Analysis
This experiment uses a laser sensor with a ranging range of 0.02m-4m, a scanning angle range of 240°, a frequency of 100ms, and an angular resolution of 0.35°. By processing the laser data, straight line features are extracted, and the lines fitted using the least squares method are merged. Then, a matching method combining point and line features is used to match the local and global maps to obtain the global map. The generated local and global maps are shown in Figures 3-4.
(1)
(2)
(3) Local map
(4) Global Map
In the process of fitting lines using the least squares method, this algorithm divides all data points into n clusters with very small intervals, fits a line to each cluster, and discards lines with large deviations. For approximately parallel lines, they are merged, and line features are extracted again. This method effectively suppresses the influence of noise on local map creation. During the matching process, by calculating the offset matrix, the local map is transformed from the robot's body coordinate system to the global coordinate system. Compared with methods that simply rely on line feature matching, this method can better reduce the errors caused during the matching process. Experiments show that this algorithm has the advantages of high accuracy, low complexity, and high real-time performance, and is suitable for simple indoor environments.
4. Conclusion
Map creation is a prerequisite for mobile robots to achieve autonomous navigation in unknown environments. This paper proposes an algorithm to create a global map by fitting a straight line using the least squares method and combining point and line features. The results demonstrate that this algorithm can effectively create maps for simple indoor environments with only straight line features.
