Søg

Robotic optimisation of IoT sensor placement


Burning Question:

“Can we build an autonomously moving sensor robot to collect environmental data from all areas of a building and then use this data to determine and visualise optimal placements for permanently installed sensors?”

Christian S. Dahl & Kristoffer S. Breuer

Aarhus University, School of Engineering

Introduction:

A proper office work environment highly depends on a proper indoor climate. A lot of tools have been developed to help control and maintain optimal conditions and comfort levels. For instance, Schneider Electric has developed a Building Management System (BMS) for solving a number of these issues in the work environment at Stibo.

The BMS is partly automated to function as a closed control loop system, and partly as a semi-automated decision support system (DSS) that requires final decision making by a building manager. The manager may change parameters in the controlling of the building’s indoor climate based on the DSS suggestions.

Clearly, in order for the BMS to function as intended, sensors must be deployed throughout the target building; such sensors can monitor any physical quantity of relevance, e.g. temperature, humidity, and CO2 levels.

Sensor placement is a critical matter, since it directly affects the BMS ability to correctly change temperature and airflow. Instead of simply installing many sensors, although a trivial, but also costly solution, it is of interest to find optimal sensor placements, i.e. such that the sensors achieve the most value in terms of information density and cost.

A closely related problem, common to a lot of buildings using a BMS, is how to convert large amounts of indoor climate data into valuable and actionable information. By finding optimal sensor placements it is possible to reduce the amount of data without compromising the information contained.

The practical part of the thesis entails using an autonomous robot by Segway called Loomo.

Loomo will be gathering data from around the Stibo building by traveling around a test area ideally without direct human interaction.

The more theoretical part will be to analyze the gathered data and reduce its dimensionality by looking into, e.g. the sample rate and the amount of sensors placed in the building. This will result in an automated solution showing where the sensors could be placed alongside a potential reduction in the amount of sensors in the individual building.

Presentation for the Stibo Systems user group in Amsterdam, september 2017

Project partners:

#Allprojects #Previousprojects