Smart Way-Finding System
Interactive in-building navigation via touchscreen and floor-embedded LED strips at TU Berlin's ZEKI / DAI-Labor
A real wayfinding installation at TU Berlin's ZEKI AI center built during the Ambient Assisted Living course at the DAI-Labor. A central Samsung touchscreen at the reception runs a Vue web app showing room buttons, a search bar and the floor plan; selecting a room sends a JSON payload to ESP32 microcontrollers that drive addressable LED strips embedded in the floor, lighting up the shortest path to the destination. The system supports several users at the same time by assigning each route a distinct color, and auto-resets after a fixed timeout.
Overview
Built during the Ambient Assisted Living seminar at TU Berlin's DAI-Labor as a permanent installation for the ZEKI AI center, replacing static signage with an interactive guidance system.
Visitors interact with a Samsung touchscreen at reception. The top section of the screen cycles through photos of the lab; the main area surfaces a button per room, with a bottom navbar for room search and an overview floor plan.
Selecting a room dispatches a configuration payload to a set of ESP32 microcontrollers running WLED, which light up the matching segment of the floor LED strip and animate the path to the chosen room.
The work was organized as a Scrum project split into a frontend and a backend subteam, with weekly planning meetings, a designated scrum master as the interface to the project owner, and Git for source control.
Architecture
Touchscreen (Vue web app) → REST/JSON → ESP32 controllers (WLED) → addressable floor LED segments per room
Tech stack
MVP features
Large room buttons sized for kiosk use, with a search bar for direct lookup and a floor-plan overview.
Addressable LED strips embedded in the floor animate the path from the reception to the selected room.
Several visitors can request routes at the same time; each route gets a distinct color so paths stay readable.
Each destination has a dedicated color, chosen so neighboring rooms are easy to tell apart at a glance.
After roughly 30 seconds the LEDs are switched off automatically so the installation returns to an idle state.
The top third of the screen cycles photos of the ZEKI lab, doubling as ambient signage when no one is interacting.
Workflows
Guidance request
- 1
Visitor walks up to the reception touchscreen and either taps a room button, uses the search bar, or opens the floor plan.
- 2
The web app maps the destination to a route, picks a free color slot, and builds a JSON payload describing which LED segments to light up.
- 3
The payload is POSTed to the relevant ESP32 controllers, which drive the floor LED segments along the chosen path.
- 4
After the guidance window elapses, a follow-up payload resets the affected segments and frees the color slot for the next request.
Project workflow
- 1
The team was split into frontend and backend subgroups under a single scrum master who owned the interface to the project owner.
- 2
Each Monday the subgroups defined their own work packages and goals for the week.
- 3
Every Friday the full team met to review the week's progress and plan the next sprint.
- 4
Code was versioned in Git; coordination ran on Signal for quick check-ins and Discord for the meetings.