Get an institutional determination before recruiting people
KOKO Research is an evolving platform for studies in living spaces. Homebot One describes experiment configuration, teleoperation, and logging, while stating that capabilities and interfaces remain preliminary. Platform access does not determine whether a proposed activity is regulated human-subjects research or has the required institutional approval.
Use OHRP's decision charts with the institution's Human Research Protection Program or IRB office before recruitment or data collection. The Common Rule applies to covered research, and institutions may impose broader requirements. Some studies are exempt, but certain exemptions require limited IRB review. Obtain the applicable determination and approval instead of assuming every home study follows the same review path.
Define who is a participant and who may be observed
A home contains people who may not be the enrolled participant: relatives, roommates, visitors, caregivers, workers, neighbors visible from a doorway, and children. Map how each person could interact with the robot or appear in audio, image, video, spatial, behavioral, or log data. These are possible study data types, not claims about what a particular KOKO build captures. Ask Homebot One and the research team to confirm the exact sensors, streams, and records used by the protocol.
OHRP explains that human-subjects research can involve interaction or intervention with a living individual or the use of identifiable private information. Identifiability is contextual. A study should not treat everyone outside the enrollment list as irrelevant simply because their name is absent. Ask the IRB how to handle bystanders, incidental collection, shared household information, people with diminished autonomy, children, and anyone who cannot use the standard consent process.
- Primary participants and legally authorized representatives, if applicable
- Other residents, caregivers, visitors, and incidental bystanders
- Research staff, Homebot One personnel, and remote operators
- People described by a participant even if they never enter the room
- Groups affected by publication, demonstration, or shared datasets
Make consent an ongoing communication process
For a protocol that requires consent, treat it as an exchange of understandable information and a voluntary decision. Explain the purpose, procedures, experimental elements, risks, benefits, alternatives, confidentiality, contacts, and withdrawal choices. Ask the institutional reviewer which consent elements, injury information, parental permission, assent, or approved waivers apply. Compensation and practical burdens should also be clear before enrollment.
Use language and formats participants can understand. Demonstrate the robot's observable states, how to pause participation, what remote operation looks like, and which activities occur when researchers are absent. Revisit consent when the build, sensor set, analysis, data sharing, or protocol changes. A household member's agreement to have a robot in the home is not automatically consent to every research use of information.
- What the robot and researchers will do in plain language
- What may be collected, inferred, retained, shared, or published
- How participants and other residents can pause or object
- Which limits apply to confidentiality and data withdrawal
- Who answers study, rights, safety, and technical questions
Align the data plan with consent and withdrawal
Create a data table for every stream and derived record: purpose, collection trigger, location, identifier, access, transfer, retention, deletion, and publication form. Separate raw sensor data from annotations, features, transcripts, spatial representations, model inputs, summaries, and research logs. Ask whether Homebot One, the institution, or a third party operates each component and which agreement controls it.
Explain the protocol's withdrawal limits before enrollment. Ending participation may stop future interaction or collection without removing all previously collected data. Ask the IRB or HRPP to review whether records must be retained, which copies can be removed, and what happens to data already analyzed or published. Do not promise complete deletion unless the study can deliver it. Include a process for complaints, unanticipated problems, protocol deviations, and security events.
Operate the home study as an approved protocol
Document the exact KOKO hardware, software, sensors, integrations, room conditions, teleoperation, staffing, training, and safe-stop method. Identify when researchers may enter the home or connect remotely. Use a pre-run checklist, participant-visible contact path, issue log, and procedure for device changes. A development update should not enter the study merely because it is available; determine whether it changes the approved protocol or consent.
At closeout, reconcile units, accounts, access credentials, data transfers, retention dates, compensation, open adverse-event or complaint follow-up, and required reports. Preserve the study record required by the institution and agreement. Publications and demonstrations should describe the tested build, setting, supervision, sample, exclusions, and limitations so a bounded result is not mistaken for a general KOKO capability claim.
- IRB or institutional determination and version-controlled protocol
- Approved consent, assent, permission, and bystander procedures
- Safety monitoring, pause criteria, and emergency contacts
- Change-control path for hardware, software, and analysis
- Closeout checklist for access, data, devices, and reporting
Choose evidence-based KOKO pilot success metrics
Frequently asked questions
Does every KOKO Research project require IRB review?
No. The institution should determine whether the activity is research involving human subjects, whether the Common Rule or other requirements apply, and whether an exemption is available. Some exemption categories require limited IRB review. Follow the institution's review process before recruiting or collecting data.
Is a signed form enough for informed consent?
No. OHRP describes informed consent as an ongoing communication process that provides understandable information, supports questions and comprehension, and preserves a voluntary choice.
How should a study handle visitors or roommates?
Map how they may interact with the robot or appear in study data, minimize incidental collection, and ask the IRB for an approved notification, permission, avoidance, or exclusion process appropriate to the protocol.
Can a participant require all research data to be deleted after withdrawal?
The answer depends on the approved protocol, consent, data state, applicable rules, and what has already been used or published. State the limits accurately and never promise deletion the study cannot perform.
Sources & further reading
From Homebot One, the team building KOKO in Fremont, California.
