Date: 2017-2018 Q2 B2.1
Course: Intelligent Interactive Products ( DBB220)
Coach: dr. J. Liang
Collaboration: Anna Gjerdrum, Buster Franken and Aidan Bundel
Expertise areas: Creativity and Aesthetics, Math Data and Computing, Technology and Realization
The LullaBear is a baby proof product designed to place in the crib and take preemptive actions to soothe a baby back to sleep, hence reducing the need of parent interventions. Thus it attempts to provide both the baby and the parents with a less disrupted sleep. The product is a combination of hardware modules inside a teddy bear that together evaluate the baby’s sleep, take actions to soothe the baby and relay information to an interface. Movement and sound sensory information is sent to a processing unit, pre-processed and then analysed by a SVM learning algorithm trained for different stages of the baby’s nightly habit. One of three predictions is made with matching actions; asleep: no action is taken, restless: a lullaby or womb sound is played, and distressed: after a certain period the parents are alerted. A reinforced learning algorithm analyses if the action taken was effective and adjusts the weight of said action to increase or decrease the chance of the action being taken. The prototype coded and built for this report demonstrates all these features and thus the goal of this project has been reached. Implementation and refinement towards a consumer product is a further implementation left for the future.
During the process, I learnt about the basics of machine learning and implementing them into a prototype for a design. Next to that, I created a digital interface that would change according to the sensor data of the prototype.