What a re-leaf: IBM's AI smarts to tell 'leccy companies when their bushes need trimming

Not satisfied with trimming headcount, IBM has turned its gaze to chopping trees that might interfere with power lines.

The technology, unveiled at Distributech 2019, is geared towards preventing power outages by identifying trees (and other vegetation) that pose a threat to overhead cables. Tree surgeons and other workers with a head for heights and a love of petrol-powered pruning implements can then trot out and have a go at the unfortunate shrubbery.

Big Blue worked with Oncor, the fifth largest utility in the US, to come up with the snappily titled "The Weather Company Vegetation Management – Predict".

The tech, built on IBM PAIRS Geoscope, produces datasets based on satellites, drones, aerial flights, IoT sensors and weather models which utility companies such as Oncor can beat into shape with the compulsory AI stick (natch).

The results can then be used to try to predict growth and prioritise areas where infringement of power lines is likely to occur, meaning a smaller risk of the lights abruptly shutting off.

It's a better idea than simply chopping down every tree that could ever possibly drop a branch on an overhead cable – a plan that came dangerously close to reality, according to an industry professional we spoke to. While wholesale felling was ultimately rejected, IBM will be delighted to hear our contact described Big Blue's tech as being better than current, manual efforts.

Those involve costly on-site inspections and meticulous record-keeping to ensure areas with vegetation are monitored. By shovelling in data gathered automatically over time, hundreds of miles of cabling can be tracked, meaning a sudden fire might just be avoided.

PAIRS Geoscope, which currently consumes 10TB of data per day, generates insights from massive amounts of geospatial-temporal data. Boffins at IBM reckon their technology has reduced the runtime of complex queries on the data sets to minutes, from weeks or months.

We were curious about how that jump in speed was achieved. IBM research scientist Hendrik Hamann explained it was all down to aggregating the colossal datasets and devising a querying system that didn't require a full-on programming team to collate the information.

"Let's say you wanted to get the average rain on all agricultural land during the growing season in Kenya. PAIRS will be able to return that information within a minute or less. Again every conventional system would require you to open up many files to retrieve the information, store the information from these files so the aggregate can be calculated."

And now the thing has been pointed at the data-points surrounding power lines.

All this assumes, of course, IBM doesn't experience outages of its own, since the tech is delivered through Big Blue's hybrid cloud. And that cloud has been a little wobbly of late. Europeans suffered a cloud storage outage last July while Platform-as-a-Service BlueMix was a bit poorly in 2017.

While neither incident was tree-related, should any vegetation look like it might overrun one of IBM's data centres, customers can be assured that Big Blue will already be on the case. Right? ®