CN105427738A - Map building method of multi-layer building based on atmospheric pressure - Google Patents

Abstract

The present invention relates to the field of map construction, in particular to a map construction method for multi-storey buildings based on atmospheric pressure. A mobile robot system is used to detect multi-storey buildings. The mobile robot includes a wheel encoder, a laser ranging sensor and an air pressure sensor. The map construction method comprises the following steps: 1. Recording data during the detection process, the data including the original range of the wheel encoder, the readings of the laser ranging sensor and the air pressure readings of the air pressure sensor; 2. Rao-Blackwellized particle Filter processing to obtain the 2D metric map of the multi-storey building; 3. According to the 2D metric map of the multi-storey building, use the principle of atmospheric pressure to segment a single-layer map. The present invention is based on an open and open-source hardware and software framework, and is easy to implement; it does not need to rely on the teamwork of multiple robots; the environment of repetition, symmetry or different architectural features will not affect the performance of map segmentation.

Description

A map construction method for multi-storey buildings based on atmospheric pressure

technical field

The invention relates to the field of map construction, in particular to a map construction method for multi-storey buildings based on atmospheric pressure.

Background technique

Constructing maps for indoor environments has always been a research problem. There are already some methods for constructing two-dimensional maps for indoor environments, and corresponding three-dimensional map construction methods are also emerging. Mapping methods for mobile robots fall into five categories: metric maps, topological maps, horizontal sensor maps, appearance-based maps, and semantic maps.

Metric maps are the most commonly used class of map building methods in robotics. Among them, the grid map is one of the metric maps. The metric map described in this article is a 2D metric map obtained by constructing a grid map. The grid map uses a probability method to represent the environment. Each grid represents the probability that the location is occupied in the environment. The larger the value, the greater the probability of being occupied, and the smaller the value, the higher the probability of being occupied. Small. Methods for calculating this probability include Kalman filtering, particle filtering, and graph-based optimization methods.

Although there are many methods for 2D map construction, only a few methods are used to construct maps of multi-storey buildings. Luca Jocchi and Stefano Pellegrini in Rome propose a method based on visual odometry to construct floor plans of multistory buildings. However, this method requires the mobile robot to sequentially build maps of each floor, and requires prominent visual features at the elevators.

SLAM (Simultaneous Localization and Mapping) approaches using multiple mobile robots have similar problems in building maps of multi-story buildings. Andrew Howard of NASA's Jet Propulsion Laboratory proposed in his paper "Multiple Robots Using Particle Filters for Instant Localization and Map Construction" to obtain a single map from each layer of robots, and then integrate them into a global map through map merging methods . Jonathan Cobben and Jamin Stewart of the University of Washington's Department of Computer Science and Engineering propose building maps of multistory buildings by locating robots where each other is on their respective maps, and then using global positioning to create constraints for merging the maps.

Michael Kager also proposed a direct solution to the construction of multi-layered building maps. Similar to Jonathan Coppen and Jamin Stewart's approach, they use global constraints to arrange the maps of each floor. Their method works on multi-robot teams, and it is based on the assumption that certain building features are common between floors. A single sensor (laser rangefinder) is used for global positioning and to generate constraints. If certain architectural features are repeated between different floors, symmetrical or dissimilar. This method does not work. Furthermore, it does not address which floor of the building the map corresponds to, so this information has to be given manually in addition.

The present invention is significantly different from the methods mentioned above, both in the way the map is generated and in the meaning represented by the map. The present invention not only obtains the global map of the multi-storey building, but also obtains the map of a single floor; in addition, the present invention does not depend on similar architectural features between floors, as long as the two-dimensional SLAM metric map can be obtained; another In one aspect, the present invention currently supports a single robot system. Based on the fact that the maps of each floor can be segmented from a continuous, global dataset.

Contents of the invention

The purpose of the present invention is to provide a method for constructing maps of multi-storey buildings based on atmospheric pressure, so as to solve the defects of the prior art that rely on similar features between floors and require the cooperation of multiple robot teams.

In order to achieve the above purpose, the following technical solutions are adopted. A method for building a map of a multi-storey building based on atmospheric pressure, using a mobile robot system to detect a multi-storey building, the mobile robot includes a wheel encoder, a laser ranging sensor and an air pressure sensor, and the method for building a map comprises the following steps:

1. The mobile robot is remotely controlled and moved by the operator, and data is recorded during the detection process of the robot. The data includes the measurement data of the wheel encoder, the observation data of the laser ranging sensor and the air pressure reading of the air pressure sensor;

2. Perform Rao-Blackwellized particle filter processing on the data to obtain a 2D metric map of the multi-storey building;

3. According to the 2D measurement map of the multi-storey building, the single-layer map is divided by the principle of atmospheric pressure.

The present invention is based on an open, open source hardware and software framework. It does not depend on similar architectural features between floors, does not require the teamwork of multiple robots, and only needs a complete robot system to obtain a 2D metric map of a multi-storey building, and then divide each floor according to the principle of atmospheric pressure. The map is segmented. The observation data of the laser distance measuring sensor and the measurement data of the wheel encoder can be analyzed through the Rao-Blackwellized particle filter algorithm to obtain a metric map of the entire building. According to the inverse relationship between air pressure and altitude, the readings of the air pressure sensor can analyze the transition of floors and segment the map of each floor of the multi-storey building.

The key idea of the Rao-Blackwellized particle filter in the second step is to estimate the joint posterior distribution of the map m and the robot's path . use observational data and measurement data , the joint posterior distribution is estimated by factoring:

Using the above equations, the path of the robot can be estimated in advance , thus constructing an environment map from the estimated path .

Atmospheric pressure (atmospheric pressure) is the pressure on objects in the atmosphere caused by the gravity of the atmosphere itself. Therefore, the change in pressure with respect to an infinitesimal height should be proportional to the gravitational force exerted by the mass of air passing through that infinitesimal layer. This relationship can be expressed as:

where P is the atmospheric pressure, z is the altitude above sea level, is the density of air, g is the acceleration due to gravity, and the negative sign means that the atmospheric pressure decreases with the increase of altitude.

Also, the ideal gas law is:

where R is the Boltzmann constant and T is the temperature. Therefore, from formula (1) (2) can get:

According to international standards, standard atmospheric pressure . Therefore, for constant gravitational acceleration and temperature, the principle of atmospheric pressure can be obtained according to the first-order integral equation (3):

Use the fact that the air pressure decreases as a sign that the robot moves to an upper level, and use the fact that the air pressure increases as a sign that the robot moves to a lower level. In addition, we assume that the pressure remains relatively stable, which means that on the same floor, the pressure difference between floors is important information between adjacent floors. Map segmentation only considers barometric pressure readings, and unlike existing methods, environments with repetitive, symmetrical, or different architectural features do not affect map segmentation performance.

Compared with the prior art, the essence of the present invention is to divide the map of each floor according to the principle of atmospheric pressure after obtaining the 2D measurement map of the multi-storey building; the present invention is based on open, open-source hardware and software frameworks Therefore, it can be easily implemented on existing systems; the present invention does not need to rely on the teamwork of multiple robots, and only needs a complete robot system to obtain the metric map of multi-storey buildings; Only air pressure readings need to be considered, environments with repetitive, symmetrical or different architectural features will not affect the performance of map segmentation.

Description of drawings

Figure 1 is a 2D metric map of the six floors above the ground obtained by the mobile robot's detection of the building;

Figure 2 shows the barometric pressure readings on the eight floors of the building;

Fig. 3 is to measure the change of atmospheric pressure and temperature on a stable point;

Figure 4 shows the atmospheric pressure readings obtained by the mobile robot during the detection process, the atmospheric pressure readings and the atmospheric pressure change rate after double-exponential smoothing;

Figure 5 is a single-layer map obtained after segmentation according to the principle of atmospheric pressure;

Figure 6 is a 3D illustration of a 2D metric map, barometric readings for individual floors, and the segmented map for each floor.

detailed description

The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

Embodiment: The experimental environment of the present invention is the Steinman Hall of City University of New York, which mainly includes offices, classrooms and laboratories. In the experiments, the mobile robot explored the public areas of the building, where the layout of the offices and laboratories on each floor differed significantly.

The mobile robot records three types of data during the detection process: the measurement data of the wheel encoder, the observation data of the laser ranging sensor, and the air pressure reading of the air pressure sensor. The collected data is processed by Rao-Blackwellized particle filter. The key idea of Rao-Blackwellized particle filter is to estimate the joint posterior distribution of map m and the robot's path . use observational data and measurement data , the joint posterior distribution can be estimated by factoring:

Using the above equation, the path of the robot can be estimated first , thus constructing an environment map from the estimated path . As shown in Figure 1, the 2D global metric map of the six floors of Steinman Hall is obtained, and the maps of the six floors are distributed on the same plane.

During the detection process, the mobile robot synchronously collected the atmospheric pressure on each floor including the elevator. Atmospheric pressure (atmospheric pressure) is the pressure on objects in the atmosphere caused by the gravity of the atmosphere itself. Therefore, the change in pressure with respect to an infinitesimal height should be proportional to the gravitational force exerted by the mass of air passing through that infinitesimal layer. This relationship can be expressed as:

where P is the atmospheric pressure, z is the altitude above sea level, is the density of air, g is the acceleration due to gravity, and the negative sign means that the atmospheric pressure decreases with the increase of altitude.

Also, the ideal gas law is:

where R is the Boltzmann constant and T is the temperature. Therefore, from formula (1) (2) can get:

According to international standards, standard atmospheric pressure . Therefore, for constant gravitational acceleration and temperature, the principle of atmospheric pressure can be obtained according to the first-order integral equation (3):

Therefore, the fact that the air pressure decreases can be used as a sign for the robot to move to the upper level, and the increase in air pressure can be used as a sign for the robot to move to the lower level. Figure 2 shows the air pressure readings on the six above-ground floors and two below-ground floors (basement and storage room) of Steinman Hall at the City University of New York.

Both temperature and atmospheric pressure have a certain relationship with altitude. Figure 3 shows the changes in atmospheric pressure and temperature measured at a stable point for 5 days. It can be seen that during the short-term experiment, the atmospheric pressure remained relatively stable within the same altitude range, but there were great differences between floors , but the change of temperature is not stable enough, so the present invention chooses atmospheric pressure readings instead of temperature readings to segment the metric map.

The order in which the mobile robot detects floors starts from the elevator on the fifth floor, and returns to the elevator after a circle on the fifth floor. Then take the elevator to the sixth floor for detection. After returning to the elevator, go directly to the fourth floor from the elevator. The same detection work has been moved to the first floor in turn. The map given in Figure 4 does not take into account basements and storerooms, only the atmospheric pressure distribution for the six floors above ground based on atmospheric pressure readings. The upper graph is the air pressure distribution graph obtained according to the original reading; the middle graph is the air pressure distribution graph after being smoothed by the double exponential smoothing method; the lower graph is the rate of change of the smoothed pressure, which can reveal the location of the elevator Location. The elevator is a closed chamber, so the heating, ventilation, and air conditioning (HVAC) unit controls the air circulation in the elevator. The effect of this phenomenon can be judged from the peak in Figure 4, that is, the peak is actually the area where the elevator is located. It can be seen from Figure 4 that the data of atmospheric pressure is discontinuous, and the data set is just divided into six continuous parts, which correspond to six floors. According to the relationship between atmospheric pressure and altitude, sort the six sections of atmospheric pressure from high to low, and you can segment the maps from the first floor to the sixth floor one by one. Figure 5 is a single layer map segmented from the metric map of Figure 1 based on barometric pressure readings.

The upper figure in Figure 6 is a 2D global metric map, the middle figure is the barometric reading corresponding to a single floor, and the lower figure shows a three-dimensional illustration of each floor map divided according to atmospheric pressure in a 2.5-dimensional space. The three-dimensional illustration is the map of the six floors above the ground of the Steinman Hall of the City University of New York constructed by the present invention.

What is disclosed above is only a preferred embodiment of the present invention, and the described embodiment should be regarded as illustrative rather than restrictive in all respects. Therefore, the equivalent changes made according to the claims of the present invention still belong to the scope covered by the present invention.

Claims (3)

1. A map construction method based on atmospheric pressure multi-storey building, it is characterized in that, adopt mobile robot system to detect multi-storey building, described mobile robot comprises wheel encoder, laser ranging sensor and air pressure sensor, described map construction The method includes the following steps: 1. Record data during the detection process, the data includes the measurement data of the wheel encoder, the observation data of the laser ranging sensor and the air pressure reading of the air pressure sensor; 2. Perform Rao-Blackwellized particle filter processing on the data to obtain a 2D metric map of the multi-storey building; 3. According to the 2D measurement map of the multi-storey building, the single-layer map is divided by the principle of atmospheric pressure. 2. The map construction method according to claim 1, characterized in that the Rao-Blackwellized particle filter of said step 2 uses the observation data obtained by the laser ranging sensor and the measured data from the wheel encoder , the joint posterior distribution is estimated by factoring: Using the above equations, the path of the robot can be estimated in advance , thus constructing an environment map from the estimated path . 3. The map construction method according to claim 1, characterized in that the atmospheric pressure principle of the step 3 is Among them, z is the altitude above sea level, R is the Boltzmann constant, T is the temperature, g is the acceleration due to gravity, P is the barometric pressure sensor reading, and P ₀ is the standard atmospheric pressure.