| Summary and Challenges |
- Currently street lighting utilizes various data such as luminosity sensor data, weather information, sunrise and sunset data to adjust the timing of power on and off. To further enhance on energy saving, the light intensity of the street lighting could be dynamically adjusted, instead of a simple binary control. With the use of AI, we aim to take into account more surrounding data than that have been currently available, such as the surrounding light intensity, color warmth, human and vehicle traffic, etc. and to adjust the dimming of street lighting such that energy can be saved and at the same time there will be no adverse impact to the public’s normal life.
- Project Challenges:
- Suitable AI model to attain energy saving
- Variety of data that can contribute to train the AI model to a satisfactory result
- Difficulties in collection of certain data that may require some other techniques, e.g. data augmentation, transfer learning, synthetic data, active learning, and semi-supervised learning to be applied
- Impact to the public, e.g. the adjustment of lighting should be in a gradient mode and the minimum light intensity should be maintained according to statutory requirements
|
