Agile Autonomous Aerial Robots

Ted talk

Video documentation of experiments with Agile Aerial Robots helicopters. Research in the context of GRASP: General Robotics, Automation, Sensing and Perception (GRASP) Lab, at the University of Pennsylvania.

Is this an example of how collision is made productive?

How about?

Inspired by Agile Aerial Robots, ‘autonomous graphical units’ could be an interesting idea to explore. Imagine a glyph that tries to finds a place on a page, but it’s not alone! ‘Hello glyphs, you know what your mission is - find your position! How can elements move in a space and change themselves. Extending the dimensions of a glyph beyond x and y to help it find it’s place on the page. How could digital objects work together when they work with/from an experience of their environment. Sensitivity and sensibility.


  • A whole class could be one single glyph. Each student is part of this glyph. How far can each letter be stressed, stretched and tortured? What glyphs would withstand these changes, which would fall apart?
  • General scale of text: You know the ‘character’. Two teams, each form a word and are dropped in a defined space. How could they be positionned.
  • Playing with agile but brainless glyphs. We can maybe replicate these spaces and observe randomisation. (white blanket, net).
  • Can we build a moveable glyph, a glyph with engines that are waiting for other glyphs to respond to? Write a program to make motors/sensors play together. This can be exercised through physical bodies too?

Mixing streetnames and -shapes


Area of Brussels depicted in Derouck map of Brussels, 1970 and in Google Maps, 2012

The first image is a detail of a map of Brussels drawn by hand. The level of precision and richness of hand-lettering seems unattainable for automatic engines. If you precisely analyse this example, you realise there are many systems at work at once. A human being has organized the overload of streetnames through abbreviation, scaling, bending and alternated placement. Discerning playfully between different solutions, all streets have been given their place in a single, legible image. The way digital maps are currently produced is surprisingly crude. It seems that the only solution is to simply leave out information when trouble arises. If you look at the second example, you understand how this can result in very poor maps.

How about?

We make an engine that can combines computing and visual intelligence? How could we mix in human intelligence; a form of automitization that can be used in different systems?


  • Compare Googlemaps and OpenStreetMaps implementations: How are decisions made? What arguments, parameters are taken into account?
  • Write-out a series of micro-solutions and decision-making process
  • Take a plotter which can visually interact with a mute map and the system needs to find the best way. Software has many micro-solutions in order to optimize the path that needs to be drawn. Could the plotter show traces of it’s decision-making process? It needs to act as a reading and writing machine.

The Text Box Model

Co-position worksession at Baltanlabs, October 2011

In digital tools, type is still virtually encased in a square box, as if it was made of lead and not bits or let alone a set of coordinates. How can we think differently about the space that a glyph occupies?

Multi-level type design interfaces

Co-position workshop, Brussels 2012

Type design is an iterative process of refining design directions. Starting from a better understanding of what happens when you design a font, we worked on a type design environment that moves more fluently between different scales of design: From single glyphs to letter-pairs and textblocks but also to move between different versions of both digital and hand-drawn sketches. Some of these features can be already discovered in Fontmatrix:

Nested optimization

Fiery-cut demo website

A software (Fiery-cut) automatically has laid out complex elements on a plane. It is a solution, not a problem. How could this work for graphical elements, could be made useful for graphical elements. This one is interesting because we wonder if this was really automatically done. It shows a range of solutions for optimizations, or displays fuzzy logic. In this image the use of blue lines and black squares is interesting and seems to help system negotiate many levels of available space. Analysis is based on more complex references than just rectangular shapes.

How about?

  • In current typesetting systems, linear ‘flow’ is the main rule and there is no concept of multi-dimensional space.
  • Could there be lay-out systems that combines different systems of optimization, taking in different parameters for‘fit?