Robots need sensors to understand the world around them. Most of the time they need a lot of them, however they are expensive. You will find that a 1/2 to 3/4 of the total cost of a robot platform is in sensors alone. Below is a helpful list of the main sensor types, including there strength and weaknesses. Most of the time you will need to have more then one sensor then merge the sensor data together. In the software world this is called sensor fusion. This merging of data is what helps the robot understand its environment and navigate it safely. For example – a Lidar system is place close to a stereo vision system. Whereas the Lidar is able to detect close objects that the stereo vision system may have missed. With the Jetson robot platform I’m looking to use as few sensors as possible but still allow the robot to do SLAM and Deep learning experiments. This ultimately will keep the cost down. In turn paving the way to home robotics of the future.
- $ Affordable for most makers
- $$ Possibly too expensive for most
- $$$ Only the finest sensors, out of reach for most unless university or a company
LIDAR: is a surveying technology that measures distance by illuminating a target with a laser light. LIDAR exists as an acronym of Light Detection And Ranging, and was originally created as a portmanteau of “light” and “radar”
- Highly accurate range detection
- Very good for localization
- Prices are dropping due to system integration of autonomous cars
- Most systems are expensive, however there are a few within reach of the maker community
- Most systems are mechanical and need a clear line of site on the robot
- LIDAR-Lite v3 $
- RPLIDAR 360° Laser Scanner $$
- LeddarTech Platform Sensor Evaluation Kit $$ *My current recommendation as the price is low at $299.00 for 16 point solid state LIDAR.*
- Sweep V1 360° Laser Scanner $$
- Hokuyo URG-04LX-UG01 $$$
- Hokuyo UST-05LN $$$ a little different then the three above as it is listed as a (obstacle detector)
Vision: Computer vision is concerned with the theory behind artificial systems that extract information from images
Sub category sensors:
- Stereo cameras use two or even three camera in a calibrated alignment to provide depth information the robot can use to get the distance to objects it sees. Needs lighted areas to work well
- IR Depth cameras use project IR light and use a filter on a camera to sense the light and based on the reflection determine the distance of a object, nice thing about these is they will also work in total darkness.
- Microsoft Kinect (old version) $
- Xtion PRO motion sensor $
- Any USB 2.0 -3.0 camera will do nicely for most $
- Point grey cameras for fast FPS $$$
Sonar: A form of range finding that is less accurate than lidar but a lot less expensive. You can cover a 360 degree arc on the outside of the robot to detect objects for the price of one fixed lidar, however there will be blind spots. Thy work with the echolocation process, sound waves that are transmitted, bounced back and are received, with the time difference used to calculate the distance of objects.
- Sonar sensors have dropped in price drastically
- They are easy to communicate too
- Not as accurate as laser based rangefinders
- Easy to fool on certain surfaces
- Need a array of them for best coverage as the beams are narrow
- They often miss things like chair legs
Internal/enviroment: This could be a sensor that measures voltage of the battery or temperature of the main processor, most of the time there is more internal sensors then external ones on robots, But they are normally passed up as they are built into a lot of the devices and modules needed to build a robot.
IMU: An inertial measurement unit (IMU) is an electronic device that measures and reports a body’s specific force, angular rate, and sometimes the magnetic field surrounding the body, using a combination of accelerometers and gyroscopes, sometimes also magnetometers.
GPS: which stands for Global Positioning System, is a radio navigation system that allows land, sea, and airborne users to determine their exact location, velocity, and time 24 hours a day, in all weather conditions, anywhere in the world. Normally the robot would need to be outside or near a large window to take advantage of this type of localization.
Are these all the sensors? NO there are like thousands of sensors, it takes knowledge and skill to pick the correct sensors for your application. If you are unsure how to connect or read from the sensors most of the time manufacture will provide datasheets and detailed guides to get you started, and if all else fails Google is your friend.
Looking for more on robots? Please take a look at the official Second Robotics recommended reading list.