We have a Turtlebot 2 that needs to be replaced. Currently, we are looking into the iRobot Create 3 as a replacement for the base. On Youtube, I found some videos where a Create is simulated in Gazebo. The simulation shows several depth sensors on the robot. I am interested in the quality of such sensors. For example, would it be possible to create a map based on such sensors (so without lidar or 3d cameras). Or do we always have to add such sensors in order to use Nav2?
Hi, I’m Alberto from iRobot, one of the developers of the iRobot Create 3.
As you noticed, the Create 3 is equipped with 7 IR sensors (see the documentation for more details Hazards - Create® 3 Docs).
The main purpose of these sensors is to implement obstacle detection algorithms, i.e. by looking at the intensity value you can determine whether you are too close to an obstacle.
Unfortunately, using these sensors for mapping is likely not doable.
The returned values depends on the reflectivity of the surfaces you are pointing at.
Through a calibration procedure you could try to define a function that converts those raw values into a distance, but that would only apply to that very specific surface.
It would also be quite difficult to ensure that the calibration is precise enough for your needs.
Having said that, my colleagues at iRobot just presented a paper at IROS titled
Trajectory-Based SLAM for Indoor Mobile Robots with Limited Sensing Capabilities (IROS 2023 Program | Wednesday October 4, 2023).
The tests in the paper were conducted on an iRobot Roomba i3, which has the same exact sensors as the iRobot Create 3.
Long story short, it is indeed possible to run SLAM using the sensing capabilities provided by the Create 3, although we don’t recommend doing that based on the IR sensors.
P.S. we have a github discussion forum dedicated to Create 3 (iRobotEducation/create3_docs · Discussions · GitHub)