Get in touch
Are you a student?
Are you a partner?
Candidates
TU Delft Impact Contest
Canopy.ag
Dutch fruit farmers are facing a labour crisis for manual tasks such as thinning, pruning and harvesting. The agricultural sector heavily relies on migrant workers, and estimations suggest this reliance will only increase in the future. At the same time, experts in the sector are sounding the alarm, warning that the once steady stream of migrant labourers could deteriorate within just a couple of years. There is an urgent need for autonomous alternatives for manual tasks such as thinning, pruning and harvesting.
Read more
LogicLayer
Industrial distributors process complex, high-stakes orders via messy "analog inbound", emails with blurry photos of rusted nameplates, 40-page foreign manuals, and hand-drawn sketches. Processing these manually means highly-paid sales engineers waste hours playing data-entry clerk just to extract basic specs. Because processing is chronological, a lucrative €50,000 engine order can sit unread while an engineer wastes 40 minutes deciphering a €500 pump request. This bottleneck costs companies major deals. Worse, generic AI cannot solve this because it hallucinates technical specs and poses massive data privacy risks when fed proprietary ERP data.
Read more
Kolibair
Emergency response lasts hours. Today’s emergency drones last minutes. Fire, flood and search-and-rescue incident commanders need continuous live visibility over hotspots, access routes, searched areas and possible victim locations, yet current drone options force an unwanted compromise. Commercial multirotors are quick to deploy, yet typically fly only 25-40 useful minutes, lose 10-15 minutes per each battery swap, require constant operator attention and rely on manual operator reporting to commanders. Military UAVs and helicopters can cover more ground, but are too costly, complex and sparsely available for regular crews. The result is broken aerial intelligence exactly when lives depend on fast, informed decisions.
Read more
Calmi Vest
Anxiety and panic don’t wait for therapy appointments. When stress spikes, people often lack an immediate, private, and effective way to downshift their nervous system—especially in public, at work, or in the wild. Current tools (breathing apps, meditations, fidgets) require attention and time precisely when attention is scarce. Weighted blankets and hugs help, but they’re not portable or “on-demand.” Meanwhile, wearables detect stress signals, but rarely translate that data into fast, embodied relief. The need: a discreet, rapid, body-based intervention that can activate at the right moment.
Read more
BRUSHLESS
Surface preparation of cargo ship hulls remains an expensive, physically demanding, and hazardous process that is still difficult to automate fully. Existing robotic systems can reduce labor costs and processing time on large, easily accessible hull areas, but they struggle with highly complex geometries. As a result, regions accounting for up to 15% of the hull surface still require manual work. These sections are often the most difficult, time-consuming, and dangerous parts of the surface preparation process.
Read more
CellaCure Technologies
Up to 400,000 adverse surgical outcomes are entirely preventable every year. Currently, 25% to 50% of malpractice cases stem from surgical errors, largely because surgeons must train on rigid, unrealistic plastics. Hospitals want to translate patient MRI scans into realistic soft-tissue phantoms, but current methods fail them. Traditional phantom labs rely on manual molding and casting, which is incredibly slow, unreliable, and cannot produce patient-specific anatomies. Meanwhile, commercial 3D printers cannot process complex medical hydrogels, forcing researchers into a 6-to-8-month "DIY" hardware nightmare that wastes up to €20,000 per project. This massive bottleneck drains budgets and keeps life-saving training tools out of the operating room, leaving surgeons underprepared and patients at risk.
Read more
Spec2Solve
Researchers with deep domain knowledge waste time translating ideas into solver-ready optimization models: choosing variables, constraints, indices, and relaxations, then debugging feasibility and scaling. This “spec-to-code” gap causes slow iteration, fragile implementations, and modeling errors.
Read more
TangentAir
Nowadays, Air Traffic Control (ATC) manages airspace sectors, which can become crowded and cause delays. The current solution to this problem is manual command issues by the ATC, which is reactive to the situations. Moreover, with the technological advancements, Unmanned Aerial Vehicles (UAVs) and large drones are becoming more mainstream, meaning that the airspace will be even more crowded in the future. It is apparent that with the expected direction of advancement, a reactive solution to crowded airspace may be insufficient, which is what needs to be addressed.
Read more
Topper Talent
Engineering students at TU Delft often work underpaid jobs that don’t utilize their full technical skills, while startups and small tech companies struggle to find affordable, highly skilled talent for short-term engineering tasks.
Read more
NeuroLearn
Traditional education follows a one-size-fits-all approach, despite students having different learning speeds, strengths, and weaknesses. As a result, students often practice skills they have already mastered, while struggling areas remain insufficiently addressed. Teachers also lack scalable tools to personalize learning for every student. To compensate, many families turn to private tutors, who are often expensive and not always aligned with the specific curriculum or learning goals. This leads to inefficient use of time, unequal access to support, and inconsistent learning outcomes. An affordable, adaptive solution is needed to personalize learning at scale.
Read more