Your robots already work.
Help us design the rooms they work in.

The problem in one paragraph

Service robots already work. They deliver in hotels, transport in hospitals, clean retail floors and move goods in warehouses. Almost without exception, the spaces they work in were designed exclusively for humans. When a robot operates in a corridor drawn only for people, it does not fail outright — it runs at a fraction of what it could do. As robots improve, the space does not matter less: it becomes the limiting factor. The decisions that determine this — floor material, corridor width, where the charging hub sits, how light is distributed, what kind of doors, whether the lift is integrated — are taken at the interior-design stage, where they cost far less than any later change. HRED proposes that this is a problem of the built environment, and that it belongs to a discipline that does not yet exist.

What the book proposes

HRED is presented as a proposed discipline, not an established one. Its components are the author's theoretical contribution and are labelled as such throughout.

• A factor model — HRED(A) = f(T, C, E, V). The autonomous capacity a space can support is treated as a function of the task families to be automated (T), the human context of operation (C), the robot equipment (E) and the physical venue (V). A structuring device, not a numerical equation: it obliges the designer to work through all four dimensions and find the incompatibilities between them.
• Nine task families, T1–T9, classified by function rather than by technology: transport, surface maintenance, manipulation, inspection, human assistance, materials handling, environmental management, security interface, exception management. Two technologically different robots that both "move things from A to B" place similar demands on the space. Consistent with the market distribution reported by the IFR.
• Six spatial zones, Z1–Z6: guest area, active service, transition corridor, back-of-house, hub, night-autonomy circuit. The hub is the only genuinely new room the designer creates; the transition corridor is hypothesised to be the parameter with the greatest impact.
• Twelve building typologies specified in detail, each with its own specification sheet, mandatory questions and commissioning indicators.
• Design parameters chapter by chapter: robot dimensions as design data, floor and wall materiality and its interaction with sensors, lighting, acoustics, circulation, furniture, network and power, human acceptance and proxemics, maintainability, failure engineering, life cycle.
• A design process and a toolkit: briefing protocol, zoning checklist, scorecard, hub sheet, commissioning and post-occupancy sheets, and a twelve-step audit protocol for existing buildings, executable by one person in half a day to two days.
• A proposed standard technical data sheet for manufacturers — including fields no commercial data sheet currently gives: the measured rotation envelope and the position of the rotation axis, the height of the main sensor plane, the maximum threshold crossable without loss of localisation, reaction latency before braking, behaviour in front of glass and mirrors.

How the book handles evidence

Every relevant claim carries an explicit status: fact (verifiable primary source), industry (institutional market data, not peer-reviewed), HRED hypothesis (the author's proposition, stated with its conditions of refutation) or open (a question the literature does not yet answer). Where there is no established science, the book declares the gap instead of filling it. Earlier drafts that contained unsourced figures were corrected and the corrections are recorded in the text. The scorecard's weights are explicitly provisional: until calibrated against field data it is a structured checklist, not a number with predictive authority.

The research programme

Eight falsifiable predictions are stated with the result that would refute each — among them: that the space explains a non-trivial fraction of performance (P1); that vertical transition is the critical point of multi-storey deployments (P2); that task family predicts sensitivity to the space (P3); that floor materiality affects localisation (P4); that furniture stability affects map-based systems (P5); that physical information anchors improve resilience under network failure without a performance penalty (P6). P1 has first published support (Mohan, Rojas & Chua, ICRA 2014: 92% versus 71% floor coverage for the same robot in a robot-inclusive versus a default layout) and P4 has independent corroboration (Borusu et al., 2024); both await replication in service contexts.

Four minimum studies, in order of increasing cost, none requiring institutional funding:

1. Inter-rater agreement. Three trained assessors apply the rubric independently to eight to ten spaces. No robots required. Precondition for everything else.
2. Floor and localisation drift. A fixed closed circuit repeated on four floor finishes; deviation between reported and real position; repeated with fiducial markers. Requires access to one robot.
3. Before/after audit in a real building. Baseline, three to five defined spatial interventions, repeat measurements. The building is its own control. Requires a willing operator.
4. Cross-building comparison. Correlational only; to be done last, not first.

An anti-circularity protocol accompanies the programme: spaces are assessed before performance data are known and by someone who does not know them; predictions are written and dated before collection; assessors are independent and their agreement is published; documented cases and illustrative scenarios never count as validation.

Asked, and offered

Asked, any one of: the site-requirements documentation given to integrators or customers; a twenty- to thirty-minute conversation with deployment, field operations or research; where possible a visit to a showroom, test floor or open client site, at my expense; deployment data (anonymised is fine) and failure cases where the space was the cause.

Offered: full attribution in the next edition and in any resulting publication, or anonymity; early access to findings, shared with partners first; a voice in the standard technical data sheet; for operators, the robot-readiness audit applied to the installation, with a written, prioritised result.

This is a student research project, not a commercial offer. The first edition is complete and being prepared for publication (Portuguese; English edition planned). I would rather be corrected than cited.