RPi Simultaneous Localization and Mapping

1. Introduction

Mapping and localization are necessary for the comprehension of autonomous vehicles about the environment they are in, allowing these vehicles to navigate without colliding with obstacles. However, a map is necessary to estimate the vehicle’s pose, and the pose is necessary for creating a representation of the map. The problem of estimating both the map’s state and the robot’s pose is called SLAM (Simultaneous Localization and Mapping). With the popularization of electronic prototyping platforms and microelectromechanical systems, it is possible to develop robots using low cost components. In this work, a prototype of a autonomous vehicle was developed using concepts of simultaneous localization and mapping to aid the navigation and to create a representation of the environment. The developed map is represented by a two-dimensional occupancy grid that provides information about the occupancy probability of specific regions in the map with a precision of centimeters.

2. Requirements

Required Python packages:

RPi.GPIO v0.6.5;
RTIMULib v7.2.1;
websocket_client_py3 v0.15.0.

Required JavaScript modules:

ws v7.0.0
jQuery v3.2.1

3. Configuration

The project folder must contain a CONFIG.json file that specifies pin connection for I/O and odometry parameters. The file uses the following structure:

{
    "pin_mode": "GPIO",
    "map": {
        "resolution": 10.0,
        "max_range": 400.0
    },
    "robot": {
        "x": 0,
        "y": 0,
        "orientation": 0,
		"measurement": {
				"front": {
						"type": "sonar",
						"pins": [10],
						"center_offset": 14,
						"orientation": 0,
						"max_range": 300,
						"min_range": 10
				},
				"left": {
						"type": "sonar",
						"pins": [4, 17],
						"center_offset": 7,
						"orientation": 90,
						"max_range": 300,
						"min_range": 10
				},
				"right": {
						"type": "infrared",
						"pins": [9],
						"center_offset": 7,
						"orientation": -90,
						"range": 25
				}
		},
		"odometry": {
			"pin": 21,
			"count_per_revolution": 50,
			"wheel_circumference": 42.73,
			"encoder_circumference": 4.08
		},
		"motors": {
			"top_left": [11, 5],
			"top_right": [26, 19],
			"bottom_left": [22, 27],
			"bottom_right": [6, 13]
		}
	},
    "server": {
        "ip": "192.168.0.100",
        "port": "8080"
    }
}

4. Unit Testing

To run unit tests, execute the script for the corresponding sensor/motor. The following files have unit tests built-in:

odometer.py
hc_sr04.py
parallax_pin.py
sonar.py
infrared.py
imu.py
motor.py
controller.py

5. Usage

Start the server:

node server.js

Start the slam script inside the firmware folder:

python3 firmware/slam.py