Recommended computer set up for MLS system

STORMBEE BEEMOBILE, MDL Dynascan, ABA Mobile Mapper, StreetMapper, IP-S2 Topcon, RIEGL VMX450, Leica Pegasus One, etc
Post Reply
pvmnt_mgmt
I have made <0 posts
I have made <0 posts
Posts: 3
Joined: Wed Feb 20, 2019 6:24 pm
Full Name: Caitlin Parsons
Company Details: IMS Infrastructure Management Services
Company Position Title: Technical Analyst
Country: USA
Has thanked: 1 time

Recommended computer set up for MLS system

Post by pvmnt_mgmt » Fri Aug 23, 2019 3:29 am

Hi LSF,

The company at which I work recently invested in a Riegl mobile scanning unit with a Ladybug 5+ spherical imaging system. We estimate that we will be collecting approximately 300GB of data a day (both point cloud and imagery), and merge each day's dataset with the previous to eventually create a comprehensive model of a city or county's roadways.
Obviously, this scale of data collection will require some intense processing power and data storage capacity. I was wondering if anyone with similar use cases/systems would be willing to share what their specs are for the processing machine/server, and maybe some of their data management work flow as well (bonus points for specific settings in RiProcess). We are preparing to purchase a new workstation and want to make an informed decision.

Thanks in advance!

User avatar
smacl
V.I.P Member
V.I.P Member
Posts: 187
Joined: Tue Jan 25, 2011 5:12 pm
Full Name: Shane MacLaughlin
Company Details: Atlas Computers Ltd
Company Position Title: Managing Director
Country: Ireland
Linkedin Profile: Yes
Location: Ireland
Has thanked: 6 times
Been thanked: 4 times
Contact:

Re: Recommended computer set up for MLS system

Post by smacl » Fri Aug 23, 2019 11:43 am

Hi Caitlin,

While I haven't used the REIGL software, I'm regularly processing MLS and image data from Leica Pegasus. My experience has been that point cloud software tends to make good use of as many CPU threads as are available and image processing is more likely to use the GPU. You'll also need plenty of memory (64gb or 128gb), fast M2 or SSD drives for currently active jobs and large cheaper backed up storage (e.g. NAS, DAS or server) for archived jobs. At a modest budget of ~$2.6kk I went for a self build based AMD TR1950X CPU, 2060 RTX GPU, 1x 1TB M2 drives, 64GB RAM, 1x8TB HDD (full specs here). On a higher budget I'd probably go TR2990X CPU, double up on both drives, 2080ti GPU and 128GB of RAM, though it is really important to talk to your software vendor to see which upgrades provide the best performance improvements. If you know you're going to have certain processes that take a lot of time and very little user input, it is also worth considering having one or more mid-range workstations accessed via remote desktop dedicated to this processing. By doing this, you can have multiple jobs at different stages in the processing pipe-line and the operator isn't left idle due to an unavoidable bottleneck (e.g. import) in the pipeline. For backup, I'd look at having automated synchronisation between your work folders and server and the server continuously backed up to something like S3 Glacier.

The above is based on my own modest office set-up, for more high-end stuff for a bigger group, I'd recommend reading some of Dennis Hirota's posts on this forum.

N.B. If you're looking at heavily automated extraction of features such as curbs, white lines and crash barriers or rails and overhead lines please have a look at SCC. (Excuse the unashamed plug :) )

Post Reply

Return to “Mobile Scanners”