Prototype to put ball in for throwing. Ball returns when the dog brings it back.
Screenshot taken from video, dog walking up to 'window' - acknowledge user.
Screenshot taken from video, throwing ball.
Screenshot taken from video, regular dog activities.
Idea: Demented elderly bonding with digital animals o. a. through interaction.
Idea: Demented elderly bonding with digital animals o. a. through interaction.
d.o.g.
D.O.G. builds on the ‘Closer to nature system’ and is designed to change the emotional state of the user and the attitude toward an animal that is not physically present. By adding this bonding experience the project aims to reduce the feeling of loneliness many of the elderly with dementia experience in their closed living environment. This is in addition to the goal of the ‘Closer to nature system’ which aims to increase the quality of life by means of offering a connection to nature and animals that is normally not available to the elderly due to their condition which causes them to lose their ability to care for their animals by themselves. The chosen animal for the system is a dog, in the outside environment, because dogs are one of the most common pets in Dutch households and many dog owners experience an intense emotional connection to their dogs. Many of the elderly also remember this connection and the feeling that comes with it.
Find out more about the D.O.G. system
By means of a gaze sensor the level of interest of user is measured (the time they spend looking and the amount of that time that they smile). This is stored for each different user. As the animal is showing various behaviours, actually the interest in the way the animal is behaving is measured. This way the system learns to recognize the behaviours of the dog that the particular user likes. To get to know these preferences for each user and to continue adapting the dogs behaviour to a users changing attitudes a learning algorithm is key. In the system ‘systematic sampling with unequal probabilities’ was used.
The system uses the learned knowledge to let certain behaviours appear more often. This is similar to how a dog learns (conditioning) in real life situations - positive reinforcement. As the user can not touch the animal behind the ‘window’ the positive reinforcement does not consist of petting or feeding the dog but rather of smiling or showing interest in the dog’s behaviour. This is similar to real life situations where the dog reacts to the human which is creating a bond, especially if the dog shows behaviour that the human likes. To increase the bonding and provide physical interaction with the dog a real ball can be thrown / rolled into a doggy door, the ball continues it’s way digitally on the screen and the dog brings it back.
More details on the ‘Closer to Nature’ project
Elderly with dementia living in a closed living environment are not allowed animals in their care home due to them losing their ability to care for their animals as their dementia progresses. But animal therapy provides a joyful experience for those elderly. As it is hard for care providers to facilitate interactions with animals or nature these are brought back by the ‘Closer to nature’ system. This system aims to offer a richer living environment to people with dementia by suggesting a connection with nature achieved through a simple tactile interaction: a water pump that pumps real water connected to a high definition screen, acting as a window, that continuously sends a live feed of a farm and which shows video feed of the goats on the farm coming to drink the water when interacting with the set up.
I aspire to improve the quality of life by offering escapism and enabling experimenting and like to include social and active playful elements; as such is the case with D.O.G. During the elective I participated in the concept design, adding the playful interaction of the ball. And Iformed a team with Vera Smoor together with whom I recorded the dog videos and categorized them. Furthermore I was, together with Vera, responsible for the programming of the concept in Processing. Where we used OOCSI to indicate and send which user was interacting with the installation from a second laptop.
OVERVIEW
Grade:
9
Client:
Closer to nature project
Course:
Embodying intelligent behavior
Coach:
Emilia Barakova (TU/e)
Competences:
Technology & realization, Math, data & computing, Design & research processes, Communication & collaboration
Key learning points:
Integration vision in initially non-fitting project, artificial intelligence, learning algorithms, object oriented programming (processing) with multiple computers, creating empathy
Deliverables:
Timeframe:
April 2017-June 2017