When 90% of a pedestrian corridor fails national accessibility standards, the people who need to act on it (municipal engineers, elected representatives, urban planners) are rarely looking at the raw data. They need a visual argument. This talk is about how we built one.
Escape the Knot is an open-source Streamlit dashboard that audits the 900 m pedestrian corridor connecting Yeshwantpur Railway Station to Constitution Circle in Bengaluru. The project takes field-collected infrastructure failure data, runs it through a physics-based simulation, and presents the output as an interactive civic advocacy tool. This talk is a breakdown of the process behind every design decision we made along the way: what we chose to show, what we chose to hide, and why those choices determined whether anyone outside our team would actually use the tool.
The talk is structured around four design problems we genuinely had to solve.
The first is the problem of making friction legible. Our core metric is the Friction Index, a 1-to-5 scale calibrated against the Active Mobility Bill. The challenge is that a number on a scale means nothing to a vendor who has been setting up stalls on a footpath for seven years, or to an MLA reviewing a project budget. We walk through our colour rubric design (grey for compliant, green through red for increasing severity), the decision to use a dark map tile to maximise marker contrast, and how the pop-up card design for each geotagged node evolved from a plain coordinate label to a structured evidence card with friction rating, GPS, status, and field notes.
The second is the problem of translating time into space. The Time Tax, the extra minutes stolen from each commuter per trip, is an invisible quantity. Nobody feels 14.2 minutes the way they feel a broken drain cover. We designed three representations of the same underlying number: a per-segment bar chart showing traversal time as a spatial gradient, a cross-persona grouped bar comparing how different commuters experience the same physical corridor, and a corridor isochrone chart showing reachable distance within a fixed time budget. The talk discusses why all three are necessary and what each one communicates that the others cannot.
The third is the problem of designing for two audiences simultaneously. The dashboard is used in two very different contexts: as a public-facing exhibit on Bengawalk's channels, and as a live demonstration tool in meetings with DULT and BBMP officials. The same visualisation has to be immediately legible to a commuter who has never seen a friction index, and precise enough to satisfy a government engineer reviewing a project approval template. We walk through the specific design choices that serve both: the expander pattern for hiding technical math without losing it, the sidebar slider for simulating remediation scenarios, and the headline metric tiles that give a BBMP reviewer the BCR number in under three seconds.
The fourth is the problem of making qualitative data sit alongside quantitative data. The project includes ten stakeholder interviews: street vendors, elderly pedestrians, a railway ticket collector, a PhD researcher, and a Swiggy delivery executive, with bilingual Kannada and English transcripts documented and embedded in the dashboard. The design challenge is that a waterfall chart and a street vendor's quote about election-season repairs occupy fundamentally different registers of evidence. The talk discusses how we structured the interview archive module to make the human narrative feel like evidence rather than anecdote.
Throughout the talk, we reference specific FOSS tools: Folium for the interactive map, Matplotlib for the static policy charts, and Streamlit for the multi-module dashboard architecture, along with the design decisions imposed by each library's constraints and defaults.
