
What should you commission before building a physical humanoid robot?
Start with a scoped digital prototype that makes the proposed character, movements and interaction understandable. Humanoid robot design brings appearance, behaviour and technical constraints together. A digital prototype can help a team review those choices before committing to physical development, but it does not certify hardware, prove autonomous behaviour or replace testing on the intended platform.
This guide is for brand teams, agencies, exhibition designers and robotics teams commissioning the human-facing part of an android experience. Mimic Androids' scope includes digital human creation, robotic face concepts, facial animation, conversational interaction design and prototype direction. Physical manufacture, sensing, control systems and deployment need separately agreed responsibilities. The practical goal is a brief that lets each specialist price and deliver a defined piece of work.
1. Define the purpose and scope of humanoid robot design

Write down one useful thing the experience should help a person do. A museum character might explain a historical object; an exhibition guide might introduce a product demonstration. Those are example briefs, not claims of existing Mimic deployments. Describe the visitor, the setting, the available time and the intended next step. This turns an open-ended request for a lifelike robot into a task that can be reviewed and tested.
Choose the output before choosing its visual style. A concept film, a real-time digital character, a projected face and a physical robot are different deliverables. They can share reference material, but they do not share all the same constraints. A rendered smile can be revised in an animation rig; a mechanical smile depends on an actual mechanism. If the project starts on screen and might later move to hardware, record that future ambition without treating it as already solved.
Make a responsibility list. Identify who supplies the character assets, dialogue design, knowledge content, speech services, software integration and physical platform. Ask which components already exist and which are still being evaluated. One studio may coordinate several disciplines, but a proposal should still say what is included, excluded and dependent on another supplier. Review Mimic Androids' capabilities against your requirements instead of assuming every service is part of every engagement.
Agree what the first review must demonstrate. For example, a short greeting could show how the character looks at the visitor, introduces its role and returns to an idle pose. That can be enough to approve an initial direction. It is not evidence that the system can recognise every visitor or answer any question. A smaller, explicit demonstration gives reviewers a fairer basis for decisions than a montage of unrelated impressive moments.
Consider the people who may not use voice or may approach from a different height or distance. Provide an accessible alternative where the setting requires it, and explain the interaction in plain language. The human–robot interaction guide offers related planning considerations. At this stage, the key deliverable is a shared statement of purpose, audience, format, dependencies and acceptance criteria.
2. Design the face, expression and conversation together

Give the character a role before giving it a personality. A helpful guide needs a different pace and vocabulary from a theatrical performer. Decide how it introduces itself, what it can discuss and how it acknowledges uncertainty. Appearance should support that role. Gather a small set of approved visual references, describe what you like about each, and note any features you do not want. If a real person's likeness or voice is involved, resolve the relevant permissions before production.
Review the face at the scale and distance at which it will be seen. A close-up promotional render is a different viewing condition from a busy public space. Request front, side and three-quarter views where they help assess the design, plus the expressions central to the experience. Consider how eyes, mouth, head direction and body posture read together. The aim is a coherent character that communicates its state, rather than detail added for its own sake.
Define an expression set around actual interaction moments: greeting, listening, speaking, waiting, clarification and goodbye. Specify whether movements will be authored animation, driven by a performer or triggered by software. An animation rig is a control system for a digital asset; it is not a robot's motor controller. Our robot facial animation guide explains this distinction in more detail, including gaze and lip sync.
Industry examples show why those disciplines need coordination. In Autodesk's Engineered Arts case study, the team describes establishing desired movement before designing the mechanisms that reproduce it. Furhat's platform description connects facial animation, neck motion, perception and audio as parts of social interaction. These are examples of other companies' approaches, not promises of compatibility with a particular Mimic asset.
For conversation, write a few realistic exchanges before adding more topics. Include an interruption, an unclear question and a request outside the approved scope. Show what the character says and how its face behaves while waiting for a response. Assign an owner for the facts it uses. A persuasive voice should not conceal uncertainty or invent a product claim. The technology workflow helps separate the character layer from the surrounding system integration.
3. Review a digital prototype against observable tasks

A digital prototype is a reviewable version of selected parts of the intended experience. It may be a rendered sequence, an interactive screen demonstration or a scene running in a chosen engine. State which one you are commissioning. The word prototype alone does not tell a supplier whether you need editable files, an executable application or a presentation video. Describe how reviewers will open it and what they should be able to change or test.
Separate visual approval from interaction approval. A face can be visually complete while its dialogue still uses temporary content. A conversation can be useful before the final character is ready. Record those states so stakeholders do not approve one layer by accident while judging another. Use a versioned review log with the scene, observed issue, requested change and responsible person. Consolidated feedback prevents several reviewers from giving contradictory instructions to different specialists.
Use representative scenarios rather than a single ideal conversation. Try a visitor who asks the expected question, someone who changes topic, a person who speaks before the greeting finishes and someone who needs human assistance. For an early prototype, these can be scripted walkthroughs. For an integrated application, test them through the actual interaction controls. Document whether responses are live, prerecorded, simulated or manually operated so that the results are interpreted correctly.
Choose measures that match the prototype. Reviewers might check whether a visitor understands the character's role, whether an explanation contains the approved facts, whether a handoff is offered at the right moment and whether the facial performance remains readable during a pause. Do not invent an engagement percentage or business return before collecting evidence. A technical team can define platform-specific timing and reliability targets when the runtime and integration scope are known.
Digital review has clear limits. A simulated room does not establish real microphone performance, installation safety or mechanical durability. Those need the relevant specialists and physical tests. The simulation and digital-twin guide provides further context. Mimic's Etisalat project contribution, by contrast, concerns character rigging and animation for an advertising film; a finished film should not be presented as proof of a live autonomous robot.
4. Prepare the brief, deliverables and budget discussion

A useful request for proposal identifies the use case, audience, visual references, target platform, existing assets and intended review date. Add the languages, essential interaction moments and any restrictions on likeness or voice. If a hardware platform has already been selected, name it and provide the relevant integration documentation. If it has not, say so. Unknown dependencies should become explicit discovery questions rather than assumptions hidden inside an estimate.
Ask for deliverables in concrete terms. Depending on the agreed scope, these could include concept views, a facial model, a defined animation set, dialogue examples, a prototype scene and a handover document. Confirm file formats, software versions, texture requirements, rig controls and whether editable source files are included. A video preview and a production-ready asset are different outputs. The same distinction applies to a conversation storyboard and an application connected to a knowledge system.
Clarify the review process alongside the deliverables. Name the person who consolidates feedback, the stages at which decisions are approved and how additional revisions are handled. Ask what happens when a requirement changes after approval. A new face, an extra language or a different runtime may affect more than one discipline. Keeping decisions traceable helps the team understand the implications before work is repeated.
Budget depends on the work required, not simply on whether the character is called a humanoid or an android. Relevant factors include asset readiness, visual fidelity, expression range, animation duration, languages, interaction complexity, integration dependencies and revision scope. Request a breakdown that separates digital production from software services, hardware, installation and ongoing support. Our project budget guide can help frame that discussion; it is not a substitute for a scoped quotation.
End the brief with a decision: what should the prototype enable your team to approve? It might be a character direction, a visitor interaction or readiness for a technical integration phase. This makes the next step specific without promising that all later work is included. Explore Mimic Androids' solutions to identify the relevant experience, then submit the brief with the assumptions that still need resolving.
Frequently asked questions
What is humanoid robot design?
Humanoid robot design brings human-like form, movement and interaction into a coherent concept. Digital character design and physical robot engineering are related but separate disciplines. A project brief should identify which parts are being commissioned and who owns the hardware and software dependencies.
Can we begin without choosing a physical robot?
Yes, a project can begin with character concepts and a digital prototype. Record that the physical platform is undecided. Later hardware constraints may require changes to appearance, movement or interaction, so do not treat an early digital design as a ready-to-manufacture specification.
What is the difference between an animation and an interactive prototype?
An animation follows an authored sequence. An interactive prototype responds to defined inputs, such as a selection or spoken request, within an agreed scope. Neither automatically proves that a physical robot can perform the same behaviour. Confirm what is live, scripted or simulated.
Does Mimic Androids manufacture complete humanoid robots?
The services described by Mimic Androids focus on digital humans, robotic face concepts, facial expression and animation, conversational interaction design, simulation and prototype direction. Physical manufacture and hardware delivery require a separately confirmed scope and appropriate partners.
What should we send when requesting a proposal?
Share the use case, audience, visual references, target platform, existing assets, required languages and target timeline. Explain the decision the prototype must support. Include known integration documentation and identify any unresolved hardware or software choices.
Can the design use an existing brand character?
An existing character can be a starting point. The team should review usage rights, source-file quality, expression requirements and suitability for the intended platform. Adapting a still illustration or film asset for interactive use may require additional modelling, rigging or animation.
How long does a digital prototype take?
A reliable schedule requires a defined scope. Asset readiness, required expressions, interaction complexity, integrations and feedback stages all affect delivery. Agree milestones after reviewing the brief rather than assuming a fixed timeline applies to every project.
How should we judge whether the prototype is successful?
Use criteria agreed before production: the character communicates its role, performs the required interaction moments, uses approved information and demonstrates a clear fallback or handoff. Record which technical conditions were tested and which still require the final platform or physical environment.
Can a digital prototype guarantee engagement or enquiry growth?
No. A prototype can help evaluate a proposed experience, but business outcomes depend on the audience, location, content, operation and measurement method. Set an appropriate evaluation plan and avoid presenting untested improvements as proven results.
Plan your next step
A strong humanoid robot design brief connects the intended visitor experience to specific assets, interactions and review decisions. Start with the smallest digital prototype that can answer the most important project questions. Keep the distinction between character production, software integration and physical engineering visible throughout the process.
Discuss your humanoid or android concept with Mimic Androids. Share your audience, target platform and visual direction so the team can help define the appropriate production scope and next step.
Contact