A challenge to solve a problem: describe the problem differently from a commissioned method
Prepare an open challenge: define the outcome, available data and assessment. Distinguish searching for a new capability from buying a known service for a company.
Article language: English

In this article
“We are looking for an AI vision system to sort parts.” A company may receive exactly what it asked for and still not know whether it needed that system. In an open challenge, the way a problem is framed also determines who will recognise it as a task they can tackle. Naming a preferred technology can close doors before anyone looks inside.
Suppose a business recovering equipment from shops wants to sort small parts into three product families of its own. The parts enter a common container after dismantling. Operators recognise some and set the rest aside for clarification. The manager wants to invite outside teams to propose another way of organising the task. It is not yet established exactly where the difficulty occurs or how to remove it.
The company could seek a solution, commission an investigation or buy a known scope of service. A challenge is one possible choice. It should have a stronger reason than the hope that many people will think for the company for free. Participants need material to work with, a way to assess responses and someone able to use the answer received.
What to leave open and what to define precisely
Openness concerns the route to a solution. It need not mean an unclear goal. The company can describe the required division of parts and what counts as an unrecognised case, while leaving different ways to achieve that result open. A camera supplier, a workstation designer and a team examining an earlier stage of the process may see different possibilities. These are candidates for investigation, not ready-made technical recommendations.
First show how an ordinary task unfolds. Where does the container's contents come from? Who identifies the parts? What happens to an item set aside for clarification? The description should let an outsider distinguish observation from the manager's assumption. “The operator has no information about origin” describes a different problem from “the operator knows the origin, but two parts look similar”.
Include real constraints too. Perhaps existing containers must be retained or difficult cases must remain available for manual handling. Explain why the condition exists. If it is merely a habit, mark it as something to discuss. Otherwise the challenge may reproduce the current way of working in a more attractive presentation.
In its 2020 RASSOR challenge, NASA invited designs for a drum to collect lunar material. The description combined collection and retention tasks with criteria including dimensions, mass and filling. The submission deadline was 20 April 2020. Open participation did not mean an arbitrary subject of work.
For the parts company, the practical lesson concerns the question's structure: participants should know what they may change and what their proposal will be compared against. We are not importing lunar requirements into a factory. Separate the outcome, environmental conditions and freedom to choose an approach, rather than ordering “innovation” with a delivery date.
Invite knowledge you cannot yet name
In a study published in 2010, Jeppesen and Lakhani analysed 166 scientific challenges involving more than twelve thousand researchers. Their original abstract describes a positive association between successful solutions and the distance between an expert's field and the problem's main field. It concerned a particular set of research challenges; it does not justify promising that any outside amateur will solve a company's problem.
Our own organisational conclusion is more modest: an industry label and a list of existing suppliers may not be enough when looking for a new approach. Describe parts sorting through the task, available information and uncertainty. A team outside the usual circle can then more easily assess whether it has a useful capability. It must still show how its proposal relates to the specific conditions.
That differs from a brief for choosing a supplier of a known service (in Polish). There you organise proposals within a defined purchase. Here you may not yet know what kind of provider you need. If scope and method have already been chosen, however, announcing an open challenge merely to obtain comparable prices would mislead participants about the contribution expected.
A public competition is not always the first step. JAXA's description of Moon to Mars Innovation distinguishes a request for information, or RFI, from a request for research proposals, or RFP. Information from the first stage helps define later topics. This describes JAXA's programme; it does not confirm that a particular call is open or that a Polish company is eligible.
A business can likewise start with a limited investigation: who knows a similar task, what data would be needed and whether the current description allows the possibility of work to be assessed at all. Such a conversation should not pretend to be a prize competition if the company is only seeking information. Naming the stage clearly saves both sides from preparing unnecessary material.
Material for a solution, not just a story of frustration
An outsider cannot see the factory floor or the gestures of an experienced operator. Prepare examples of parts and explain how the company currently identifies their product family. Mark cases you cannot assign unambiguously. Hiding them might make the task look better but would make responses harder to assess. There is little value in giving the problem a photo shoot that hides its difficult angles.

Separate material available at the start from what can be shared only at a later stage. Decide internally which data and photographs may be disclosed and who approves the scope. Before inviting work, agree with the appropriate person on confidentiality, rights to the material and conditions for later use of a solution. The word “challenge” does not settle these matters.
Do not expect a participant to guess what the organiser considers a correct division. If examples contain inconsistent labels, clarify their meaning first. Perhaps the missing definition of the families is itself part of the difficulty. The team may need to organise its own catalogue before inviting anyone to design a device. That is a useful finding even if it does not immediately produce a spectacular demonstration.
In the factory sample, one part has two different labels depending on who prepared the earlier list. The coordinator explains that the families are meant to support subsequent use, not reproduce names from old shops. She corrects the label and keeps a record of the change. Otherwise, participants might faithfully reproduce inconsistent data and then be told their solution was wrong.
Another part is hard to assign without knowing where it was dismantled. The company marks that absence in the task description. It allows a proposal to preserve the information earlier in the process, but does not automatically award that approach victory. It still needs to establish who would record the information and whether it will be available in ordinary work. Moving the point where a problem is solved may also move the effort to another team.
Checking a proposed solution against samples connects with defining the object of inspection in a vision system (in Polish). That article moves into a specific application of a particular technology. In an open challenge the question comes earlier: is that the capability we need, and what would allow us to test it? Do not settle the answer through the challenge's name.
Assess the answer to the problem, then readiness for further work
Write down what participants should submit at this stage. A concept description, the result of a limited trial and a finished device are different contributions. Requiring a complete solution in the entry form can exclude people with useful knowledge who cannot do that work without agreed funding and access to the site. The company must also count its own work in preparing and assessing submissions.
Before entries open, decide how questions will be handled. If one person asks whether the collection method may be changed, the answer changes everyone's understanding of the task. Share that clarification with all participants so they know the change is allowed. An account of an entrant's original proposal, however, should not automatically enter a public list of questions. The coordinator must distinguish the two types of material.
A participant may also request an example the company does not have. Record the absence plainly. If the promised assessment cannot be carried out without it, revise the stage's scope and inform participants before they submit. Otherwise the jury will compare confident statements with cautious authors, rather than their answers to a common question.
When comparing solutions, check whether they address the same version of the problem. One proposal may identify the family at the end of the process. Another may preserve that information during dismantling. Before awarding points, establish whether both fall within the permitted scope. Otherwise a better idea could lose because it solved the problem earlier than the presentation template anticipated.
Information about the author can return in a separate step. Assessing a concept without the author's name (in Polish) concerns what an assessor should know on a first reading. Choosing further collaboration requires establishing who can carry out a trial and what they need. An interesting proposal is not proof that an entire team is ready to implement it.
If different organisations submit together, an agreement on a joint offer (in Polish) is also useful. The organiser needs to know the actual division of work and who can answer questions about each part. A list of logos will not explain who will come to the factory, prepare material for checking and discuss an unresolved case.
When looking for teams in Korea, Japan or China, Aeneia Oriens can help identify the technological environment and potential partners from the stated problem. The enquiry should distinguish a manufacturer of a finished device from a research team and an integrator. Each may answer a different part of the task.
Someone must take the next step after the challenge
The organiser should know who will receive and use the result. At the equipment recovery company, it is not enough to announce an interesting proposal was selected. The person responsible for the parts workflow must arrange access to examples, operator participation and the scope of further checking. Without that person, the challenge ends with a prize while the problem returns to the container with the unassigned parts.
The selected direction can then be examined in a limited business experiment (in Polish). This is another decision: which assumption needs a trial, and what will its result change? Winning an assessment of proposals is not the outcome of that trial. Record what has not yet been tested so that a ceremonial conclusion does not become unintended consent to full implementation.
Before announcing the challenge, read its description as someone who has never been inside your company. Could they identify the required result, available material and limits on freedom? Do they know what to submit and how it will be assessed? If the answer depends on a conversation with one manager in the know, complete the description first. A good partner should have the chance to recognise the problem before being asked to solve it.







