Rethinking social interactions in AI language models with a non-humanoid robot and the emotional ranges it can convey.
Experimenting with consiousness in machines, ROBOTI is an establishment of connection between a large language model and the physical world, creating an embodiment and affording the model possibilities of physical experience. ROBOTI, is quite literally a shy robot, and the goal of the users interacting with it is to create a connection and make yourself approchable so that he feels more comfortable.
Collaboration: Luca Busby, Elias Müller, Aisha Monsar
In this highly digital age, social interaction is a crucial part of our daily lives. With ROBOTI, we wanted to have a tangible and material input through with people could interact and also have the ability to change to shape and form. Two key aspects we focused on were its emotive capabilities and kinetic motions.
Each movement of the robot is assigned to a specific emotion state. Initially, he is quite afraid and gives noise feedback similar to distress calls. With the voice input of users, it recognises emotions and moves towards you if it likes you or turns the other way around if he dislikes you.
ROBOTI relies on its simplistic, organic and fluid look. We used stretchable fabric to achieve this. We realised that simple mechanical motions can convey a lot of emotions, and used this as a base point to map out an emotion scale based on servo angles, servo speed, voice and capacitive touch inputs.
Connecting Inputs to ChatGPT
Prototyping different forms
Eventually reaching Baby Yoda form
Overall Architecture
Emotion scale
Tech Specs: