View Full Version : [Question(s)] Hokuyo laser Range finders

02-08-2013, 04:07 AM

I'am looking to buy a Hokuyo UBG-04LX-F01 laser range sensor for static (like walls) and dynamic(like people) obstacle avoidance in a indoor environment.

There specs says that Scan Time is 28ms.. what does this means??

It has 240 deg viewing angle. so does it scans whole 240 degrees within 28ms?? or is it just scan a one point within 28ms??

(If latter is the case full 240 degree scanning time would be = number of scanning points within 240 deg * 28ms)

following is the manufacture specs for this product.

02-08-2013, 04:15 AM
The scan time is how long it takes to make a scan -- that is the whole 240 degrees and many many data points. Low end lasers do 5-10 full scans per second, higher end devices like this are typically 30-40hz update rate. You will get about 680 datapoints per scan with that laser, or about 24k datapoints per second....


02-08-2013, 04:23 AM
Thanks mate.

Following is obtained from their data sheet. It says output one pulse in every scan for 4ms. So i thought 28ms is for just one data point.
"Output one pulse in every scan for 4msec."

Have you use this?? is it reliable in a dynamic indoor environment??

02-08-2013, 05:16 AM
That pulse is not for the data, that pulse is a sync output that tells you when the laser is in a particular spot (which might be useful if you have say a tilt motor and need to synchronize the data to the motor angle).

I've not used the UBG models, but it seems to fall somewhere between the URG-04LX-UG01 (of which I have dealt with many) and the UTM30-LX (of which I have dealt with a few). The UTM30-LX is a much longer range sensor, and is the one used on the Willow Garage PR2. It pretty much sees everything within 30m that is not glass. The URG is about 1/5 the cost, but sees only about 4-5m. It might miss dark surfaces, etc. Overall, it works fine for a robot navigating in smaller environments using ROS's default navigation stack.

If you are trying to localize/map/navigate, the overall success of the laser deployment really will depend on matching the laser range to the environment size. If you're just trying to avoid running into things, any of the Hokuyo lasers will probably be fine.


02-08-2013, 08:15 AM
Nice blog mate..... M really happy to have a advice from some who knows robotics very well....

As u mentioned my requirement is to localize/map/navigate... But i thought 30m would be too much. that's why i went for 4-5m one. For the localization i was hoping to use encoders and a good IMU....

What do u mean by "If you are trying to localize/map/navigate, the overall success of the laser deployment really will depend on matching the laser range to the environment size."

4m within 240deg is considerable area. But do u think this will not work well in large environments??

02-09-2013, 08:56 AM
It really depends on how large "large" is, and what you are trying to do.

For *building* maps, the reference "off the shelf" SLAM method is probably the ROS package "gmapping". This is heavily dependent on your laser always seeing some sort of (recognizable) feature. For instance, if you want to drive the robot down a hallway, you want to be able to see the *end* of the hallway (the walls to your sides aren't *recognizable* features). Thus, if your hallways are <4m long, a 4m laser works. If you want to map in a hallway that is 6m long, it may not work so well if the walls don't have some sort of recognizable features (think support posts, or alcoves).

For *localization* against an existing map, the reference "off the shelf" method is likely AMCL, which has a ROS package available. Here, range deficiencies can be made up for by better odometry or IMU as the algorithm doesn't just assume the laser is always right (like gmapping). Still, if you want to localize in large open spaces (huge conference rooms devoid of stuff in the middle) and your odometry isn't perfect (and it rarely is) a long range laser may still be necessary.

It really all comes down to what environment you are operating in, what you intend to achieve, and what software you intend to use -- unfortunately, one size does not fit all. I'd suggest highlighting exactly what you want to do and soliciting feedback on that *exact* set of use cases as, unfortunately, the number of potential combinations is pretty much unbounded.


02-10-2013, 02:08 AM
It was really helpful. thank you very much.