Maximilian Brandl · Independent R&D

Gefühlswelt

Nov 17 – Feb 18, prototype

Can an interface be controlled by nothing but handwriting and shapes?

about mental health: journaling helps. but most people we asked felt they can’t draw, and still wanted to express their feelings.

so:

• hide the tech. a normal journal + a pencil, like magic paper
• pencil → ML → shape recognition → the journal reacts

no buttons. no menus. no app feeling.

• the shapes worked kind of well
• people drew a lot of cats

w/ Alessa & Janett

Photos: the prototype; draw a feeling

no talent needed.

ParcelKey

Apr – Jul 18, done

problem: the "last mile"

What if your parcel finds you?

like the owl in Harry Potter: the letter finds you.

• one glowing keychain per online order
• it glows when the parcel is close (beacons, bluetooth + the driver’s GPS)
• your ordering becomes tangible (CO₂!)

should be tiny, keychain size. the prototype is… bigger.

more tech in every parcel ?

verdict: a bad idea with good parts. a key for every order makes reckless ordering tangible. I liked that.

w/ Domenika

Photos: it glows when it’s close; inside the key

never miss a parcel again.

Talent Award, nominee 2018

The Force

Jun – Jul 18, prototype

(very dramatic title. I know.)

ARKit + CoreML were just announced. I wanted to push them and see what’s possible.

Can I move things in AR with plain hand gestures?

• a small gesture classifier, on the phone
• grab · release · aim · shoot
• hardest: enough training data, with different backgrounds

impressive. but you have to hold the phone: a barrier between me and the magic.

note from the future: a looong time before Vision Pro

learned Swift + ARKit in an elective @ h_da w/ Manu from Apple

Photos: move it with your hand *; the title card

* not affiliated with Disney or Lucasfilm. the little soldier is censored, to be safe.

MiniControl

Oct 18 – Feb 19, done

Can I drive a toy car with my mind?

BCI = brain–computer interface

• my old RC Mini Cooper: new motor control, lights, horn, a camera, a Pi
• an API, so I can drive it with an Apple Watch or my BCI framework

Watch / BCI → API → Mini → drive!

• watch: the gyro, as if I hold an invisible steering wheel. worked well
• BCI: after 20-30 min of training it felt like telepathy. after lunch: gone

tried thinking words, emotions… in the end: imagining the movement.

Photos: test drive; the test pilot

Bond used a phone. I used my mind.

BrainFramework

Oct 18 – Aug 19, done

BCI-BRAINFUCK BRAINFRAMEWORK

renamed, so certain parties felt more comfortable using it. (I’m just bad at naming things.)

What if students could make games + apps that react to their mind?

the Animation & Game program @ h_da had a BCI. unused: the docs were too hard.

• so: a Unity plugin + a JS framework for the Emotiv EPOC+
• deal w/ Prof. Martin Leissler: I get time with the headset, the students get the plugin
• goal: from headset to a mind-controlled game in one afternoon

→ for my MiniControl: yes.

• Unity (done)
• JavaScript (done)
• a better name (done)

Photos: live brain data; the Unity plugin

plug in. think. play.

reply to the NASA mail!!

Aiiot

Dec 18 – Nov 19, prototype

Can anyone train their own intelligent space?

smart homes back then: for nerds (Node-RED…) and not really smart. but everyone has very own ideas.

• train your own image + sound recognition, on a Raspberry Pi
• a super simple node interface, for non-tech people too
• a few samples, one click to train

cloud local AI. already back then.

a home should stay private. by design.

w/ Philipp

Photos: the room learns; first prize!

teach your room.

Talent Award, 1st prize 2019

ThingsCon exhibitor, Rotterdam 2019

SEE Conference, exhibition · 2019

at the exhibition people made up their own rules: book in hand → reading mode. baby cries → soft music.

Symbios

Apr – Jul 19, done

an experiment in human ↔ non-human interaction

What if a forest felt your touch, and answered?

• plants = the interface: sensors + LEDs on the plants, projection mapping
• touch a plant → it glows
• but every touched spot slowly goes dark
• a Kinect watches how people behave

the idea: it’s okay to step back. if everyone touches less, everyone gets the show.

most people enjoyed the short glow, but didn’t notice that all the touching darkens the installation over time.

my phase of odd touch surfaces → later: water (Atlantis), a shirt, a couch…

w/ Hannah, Lauritz, Nadine, Silvio

Photos: the forest; what the Kinect sees

touch less, see more.

Atlantis

May – Jun 19, prototype

Can a bowl of water become a touch screen?

away from the black mirrors: after plants, what about water?

• kitchen foil + water + an Arduino: a frequency sweep
• turn the signal into a picture → image classifier + transfer learning: little training data

• signal visible (done)
• classifier works… roughly (done)
• understand what the signal means (not yet)
• keep the notebook dry (not yet)

roughly = how much hand, how many fingers. the trick: swept frequency capacitive sensing (Touché).

w/ Philipp

Photos: the whole idea; hand in the water; kitchen foil = sensor

it worked. roughly.

Tangible

Nov 19, abandoned

Can a couch be a remote control?

plants, water… next: clothes + furniture. I saw Project Jacquard: could I build that myself?

step 1: a touchable shirt

conductive yarn, sewn by hand into a shirt of mine. it worked! a very low-res grid, one thread per x and y: touch + rough swipes.

step 2: the couch

next week never.

the sewing took forever, then the semester started.

Photos: touch → it reacts; sewing a sensor

maybe one day.

later: today I’d use a sewing machine.

Entangled Interfaces

Apr – Jun 20, done

If there will be a post human-centered design, what will this mean for the future of designers and the design of interfaces?

• post-human-centred design: theory → practice
• an interface between many actors, human + non-human
• designers as mediators

w/ Philipp

a concept sketch:

Photos: actor network

who else is in the room?

Entangled Interfaces: The Design of Post Human-Centered Interfaces M. Brandl, P. Kaltofen · 2020

→ read the paper

TU Delft, guest speaker, More-than-Human Design · 2021

ThingsCon speaker, Good Things → Good Systems

next: build it as a room →

→ ThingsCon zine

the paper in one line: we need tools for complexity.

Butterfly in the Room

Jun – Aug 20, prototype

the paper (Entangled Interfaces), turned into an installation.

Can we see what one design decision sets off? and who pays for it?

• one unnamed IoT device on a table
• pick cards = make design decisions
• the hidden network of consequences lights up

inspired by Kate Crawford + Vladan Joler, "Anatomy of an AI System"

is that our job, as designers? if not, whose?

covid. prototyped in our all-white cellar. the projector heated it up in minutes.

w/ Philipp

Photos: the installation; ThingsCon 2020 talk

the user is not the only one affected.

Talent Award, nominee 2020

Mobility Design Guide

Feb 21 – Apr 22, done

How does research become a tool for planners?

• an interactive guide for the future of mobility
• for planners, politicians, architects, designers
• from big visions to concrete examples

my first contact with the HfG Offenbach. met lots of nice people → later OIMD → later Immersity Lab

w/ Jan, Philipp, project-mo.de & OIMD

Photos: early prototypes; the guide

research you can click.

published: A. Krajewski, S. Reitmaier, K. Vöckler, P. Eckart: Mobility Design Guide: Making Future Mobility Tangible and Experienceable in: Mobility Design: Die Zukunft der Mobilität gestalten, Bd. 2: Forschung. Berlin 2023, pp. 58-69

hardest part: a CMS for a 3D world. we had to push WordPress past its limits. it ate the time for performance, which could be better today.

Monsterra

Jan 22, abandoned

What if a houseplant behaved like a cat?

animism, as an experiment in human ↔ non-human interaction. a "home-scale" Symbios.

• we humanise cats + dogs. why not plants?
• thirsty → it comes to you and annoys you
• needs sun → it drives there
• how do we live with an autonomous room-mate? and don’t they have agency already?

never finished building: no time, then my Master.

Photos: concept; a rendering

the real one still lives. struggling sometimes.

Oberwesel I

Apr 22 – Mar 23, done

a museum in a really beautiful town, full of small magic · w/ Philipp, Veronika & Alexander @ 21ct

Magical Stadtbuch

Can you leaf through a town’s centuries?

an enchanted fairy-tale history book!

• old sketches of the town, brought to life
• turn a page → live through Oberwesel’s history

Chatting Portraits

What if you could chat with the people in the paintings, and hear about their time?

• the plan: let them talk. GPT wasn’t good enough yet
• so: scripted chats. no hallucinations, full control of all four stories. honestly better

Photos: the book; so does she; she writes back

they write back.

Oberwesel II

Apr 22 – Mar 23, done

same town, same team. now: the Rhine

Rhine Journey

Can an old ship’s wheel sail again?

• the original steering wheel, enchanted
• a mini game: drive along the Rhine
• and learn about it on the way

Rhine Underwater

What lives down there?

• the river’s rich underwater life: plants + animals
• projected into a walk-in cellar
• with an app, you interact with the projection

Photos: take the wheel; fish, projected

no diving gear needed.

Pointer

Nov 22 – Jan 23, prototype

Can a camera become a sensor for where someone is pointing?

a learning project: write my own model in TensorFlow, from scratch.

• my own tool to generate the training data (my arm got heavy)
• MoveNet pose data → my own model
• it predicts the target on the wall: good enough to pick areas

ML is necessary ML was the point.

could this replace the Kinect? no. wrong approach. but I learned a lot.

Photos: the model; training data

just a camera.

→ next: a whole room

Stress Copilot

Mar – Jun 23, done

most people don’t understand their own stress. so: measure it, and learn.

Can you and a self-learning system figure out your stress together?

• like Pomodoro, on steroids
• everyone has their own focus pattern
• when is the best moment for a break? which kind?

a self-learning feedback loop:

EEG + reinforcement learning

simulated users of different types: the algorithm found a sweet spot for each. in theory.

Photos: the loop; Emotiv EPOC+

your break, not after 25 min. but when you need it.

tested on me + two students in their Master thesis: real stress. the catch: the headset caused headaches, which forced breaks too. so: the tool, or Pomodoro?

Interactive Room

Oct 23 –, abandoned

What if any room could easily become a spatial UI, without a headset?

• for future museum + installation projects
• analog interaction: no device, no screen
• a framework to tell stories through interaction

the Pointer, continued. bigger, and without the pointer model.

MoveNet + Unreal / Unity: easy to wire up. then: no time, and no paid project to fund it (yet).

Photos: first render; pointing → UI

every room is a screen?

Interactive Table

Oct 23 –, abandoned

Can I build a bridge between a projection and what my hands do on a real table?

• the interactive room, but smaller: objects on a table
• an experimental framework
• again museums + installations: stories told through interaction

Photos: first quick prototype

no headset!

a cup = a button? a book = a map?

paused: no time, and no paid project to fund it (yet). still want to finish it one day.

Puzzle IDE

Nov 22 – …, abandoned… for now

How do we keep a mental map of our problems, when AI writes most of the code?

• a graph-based, infinite canvas for solving + exploring problems
• code, systems, ideas = puzzle pieces (the puzzle = the problem)
• connect them in one universal, visual language?
• translate Rust, Lean proofs… into something everyone can see + explore
• easy like Scratch, because it’s visual

everything is a graph?

complex systems too: map them, share them as a visual language, combine them with other people’s systems.

grew into an eierlegende Wollmilchsau: a moodboard + canvas prototypes. abandoned for now, not forgotten.

one tool for all kinds of problems?

does AI make us lazy? we’re problem solvers. maybe we need a tool to think with the tool.

mTOOLS + mSERVER

2026 – now, running

What happens when anyone can build their own tools?

experiments in serious "vibe-coding": home-cooked software for

one family + one small team my whole digital life

• native apps: notes, books, sync, keys. offline, in plain files
• servers that look after themselves
• a Rust core per app, thin UIs

frameworks back to vanilla: Rust, HTML, CSS, JS

what does this mean for my profession?

• just another tool to express my ideas faster
• a way to push the limits
• how do I keep my brain thinking, and understanding?

trust?

focus?

home-cooked, not mass-produced.

in daily use: all of them. my reader + to-do app fit me perfectly. family: distribution still unsolved.

Now: Immersity Lab

2023 – now, founding

Can we test a space with real people, before it exists?

Immersity Lab. first: R&D in the research group at the HfG Offenbach. now we’re founding it (Oct / Nov 26).

• VR replicas of planned spaces
• real participants: gaze, movement, wayfinding
• AI-aided analysis → evidence, not assumptions

why I’m in:

build tools to evaluate + create better spaces. spaces that work, where people feel good, with less frustration, and where their opinion is seen.

immersitylab.com