Personal fabrication has mostly focused on handheld tools as embodied extensions of the user, and machines like laser cutters and 3D printers automating parts of the process without intervention. Although interactive digital fabrication has been explored as a middle ground, existing systems have a fixed allocation of user intervention vs. machine autonomy, limiting flexibility, creativity, and improvisation. We explore a new class of devices that combine the desirable properties of a handheld tool and an autonomous fabrication robot, offering a continuum from manual and assisted to autonomous fabrication, with seamless mode transitions. We exemplify the concept of mixed-initiative physical sketching with a working robotic printer that can be handheld for free-hand sketching, can provide interactive assistance during sketching, or move about for computer-generated sketches. We present interaction techniques to seamlessly transition between modes, and sketching techniques benefitting from these transitions to, e.g., extend (upscale, repeat) or revisit (refine, color) sketches. Our evaluation with seven sketchers illustrates that RoboSketch successfully leverages each mode’s strengths, and that mixed-initiative physical sketching makes computer-supported sketching more flexible.

We present Print-A-Sketch, an open-source handheld printer prototype for sketching circuits and sensors. Print-A-Sketch combines desirable properties from free-hand sketching and functional electronic printing. Manual human control of large strokes is augmented with computer control of fine detail. Shared control of Print-A-Sketch supports sketching interactive interfaces on everyday objects – including many objects with materials or sizes which otherwise are difficult to print on. We present an overview of challenges involved in such a system and show how these can be addressed using context-aware, dynamic printing. Continuous sensing ensures quality prints by adjusting inking-rate to hand movement and material properties. Continuous sensing also enables the print to adapt to previously printed traces to support incremental and iterative sketching. Results show good conductivity on many materials and high spatial precision, supporting on-the-fly creation of functional interfaces.