What is common between a Waymo/Uber Self Driving car and the latest iPhone. The LiDAR! And that is what we hope to delve deeper into over the next few paragraphs.
LiDAR is an acronym for Light Detection and Ranging. Put simply, it is a sensor that estimates the distance to an object by shooting a ray of light and measuring the time taken by the reflected ray to return to the point of origin. You can find more details about the history and the technology behind LiDARs here. LiDARs come in various form factors. Velodyne, Sick, Ouster, Hokoyu, Slamtec, Pepperl +Fuchs are some popular vendors of LiDARs.
A mobile robot needs to have a 360-degree view, not just what lies ahead of it. To achieve this, it will need to fire rays in all directions and capture the resulting reflections. Instead of doing this with multiple LiDARs, it is a common practice to spin a LiDAR on its vertical axis. These are the Spinning LiDARs that are commonly used in mobile robots. A 3D LiDAR is an extension of the 2D LiDAR. It can be visualized as multiple 2D LiDARs that are stacked together.
In order to achieve autonomous navigation, mobile robots need to gather information about what is around them. Such information is used to:
1. Create a map -
· To know where it is in the room. This is called localization. If the robot is localizing while simultaneously creating the map of the place, it’s called ‘Simultaneous Localization and Mapping (SLAM)’. LiDAR (for SLAM) is not needed if an absolute positioning technology like Indoor GPS is being used.
· To figure out optimal paths from start to goal location. This is called path planning.
2. Locate obstacles and go around them.
3. Move around safely
LiDAR is one of the most robust and easiest ways to do all three of them.
How to choose a LiDAR:
1. For SLAM
In extremely simple terms, the LiDAR-based SLAM works by matching the current scan with previous scans or map/scan-based models. Therefore, the scan at every point of the time should contain features (walls, furniture, or other physical objects) that can be identified.
A LiDAR used on an indoor office robot should have a range that is at least 20% more than the distance between the parallel walls of the rooms the robot is operating in. Now for warehouses, this is hard to achieve, therefore the range of the LiDAR can be decided based on the distance between shelves and machines. The range of the LiDAR should be enough such that at each point it should be able to capture at least one static/permanent component in that warehouse. If a robot with a LiDAR that has a range of 10 meters is left in the middle of a stadium, it will fail to identify permanent features. This will lead to large errors in the robot’s position estimation.
This page does a good comparison of different LiDARs in the market.
2. For obstacle avoidance
The range of the LiDAR used for obstacle detection depends on a host of factors: Obstacle horizon, speed, load, inertia, and dynamics are some of them. , For a robot with large inertia, the obstacle horizon should be large too. A short obstacle horizon for a robot with large inertia can cause collisions. The range of the LiDAR should be at least 20% more than the obstacle horizon.
3. Safety Lidar
LiDAR is quite a popular option for safety sensors in Autonomous Guided Vehicles and Autonomous Mobile Robots. Most of the time it is a mandatory requirement for industrial mobile robots. More on this here.
Mowito provides a brilliant trajectory planner that can avoid dynamic obstacles and operate at frequencies as high as 50Hz using the LiDAR data. If you are building an AMR, feel free to write to us at aditya@mowito.in and we would be glad to work with you.
