TIL ティ・アイ・エル株式会社

北大発認定ベンチャー企業
JP EN

AI Road heating optimizer

In collaboration with: Hokkaido Gas

The AI road heating solution uses deep learning and image recognition to recognize various snow conditions.
With better control over the heat source, implementation of this solution can save a significant amount of energy in areas with snowfall.

What is road heating

Heating devices are installed beneath the surface of parking lots, roads and more. The heat sources help to control the amount of snow on the surface, preventing accidents, while eliminating the need for costly snow plows.

Conventional snow sensor control

Maintenance costs often increase due to wasteful operation of heat sources

These systems are installed outdoors to detect surface moisture. (Such systems can not directly observe snowfall on the surface)
As a result, the systems must run the heat sources longer to ensure adequate snow removal, and can not distinguish between rain and snow.

Flow

Thanks to constant data accumulation stored on the cloud, the AI Road Heating Optimizer is able to achieve granular control over road heating, according to the current snowfall conditions and accumulated snowfall on the surface.

Step.1

A camera is setup to monitor a part of the surface.

Step.2

Images of the surface are taken at regular intervals.

Step.3

The presence or absence of snow is determined by the neural net.

Data accumulation

Allows optimal control of the heat source according to the snowfall on the road surface

Operations

The operation status can be confirmed and changed via the internet.

Controller with built-in camera microcontroller

  • Raspberry Pi 3 Model B+ :
    Accumulation of snowfall is determined by the deep layer
  • Raspberry Pi Camera Module :
    Takes images of the road
  • LTE Modem :
    Monitors the state of the device through an Internet connection

Image recognition AI

Recognizes the snowfall conditions and outputs a signal to operate the boiler at the appropriate time

Original image

ResultSnow Not snow

Experiment

The operational time for a boiler (= Fuel cost) used to heat a hospital parking lot in Sapporo City was reduced by 29.0%.
* Reduced energy consumption by up to 85.7% by location

Accumulated snowfall Boiler operation Snowfall sensor reaction
Private home Parking 53.9% Reduction
Hospital visitor Parking 46.7% Reduction
Apartment Parking 35.7% Reduction
Hospital staff Parking 54.3% Reduction

Other applications

Roofs that require
snow removal
Sprinkler
control
Solar Power
Panel Maintenance