The aim of this project was to make a machine which is controlled by human movements and through this facilitate the healthy and ergonomic body postures. The theoretical background for this was Schusterman’s (1999) Somaestatic which is the “Somatic acknowledgment that there is no separation between mind and body and connects the self with all of these processes” (Höök 2018:xvii). In this respect, the controlling of the robot promotes to practice ergonomic body postures, like strengthening the back or lift the overarm above the shoulder.

The Somesphere is driven by a wireless Arduino, two DC motors and one servo motor. The DC motors give the speed, meanwhile the servo motor adjust the iron ballast in the middle which responsible for the direction. The outer shell and the engine holder box were produced by additive manufacturing techniques from PLA filament.

The spirit of the device is the Touchdesigner software and the computer vison centered machine-learning framework plugin, the Mediapipe, which could estimate a digital skeleton by an AI based algorithm. With this method from a live video input the distance of user’s mouth from shoulder is estimated which controls the speed. For direction the relative difference between the height of the right and the left elbow is responsible.

The first prototype had difficulties of turning the outer shell because it was too heavy and the used DC motors had not enough turning force. To test the software and estimation method the shell was replaced by lightweight wheels from reused plastic. This solved the problem but limited the shape and the movements of the device.

To keep the spherical shape the weight of the shell should be reduced by manufacturing it using a random Voronoi cellular structure (see more Rokicki & Gawell 2016, Feng et al. 2018, Zhang et al. 2021) and/or stronger motors even brushless (BLDC) motors would solve the speed and turning force problems.

In the long term two overlapping directions are seen for further work. 1) developing the human body driven controlling mechanisms by researching the ergonomic movements and postures and find the best markers which captures the most the desired gestures. 2) shape driven: experiment with the shape and the controlling hardware mechanisms in a less gravity dominated space like floating in water or levitating it in the air. This direction would also broaden the possibility to translate human movements into different controls.

The hypothesis was that these kinds of movements, even first need a high self-reflection and body consciousness, but after a while, when these are already used to, it may help incorporate the movements. Same as cars are driven without focusing on the control mechanism