Luna Methods, a Dublin-based startup, has constructed one thing which may truly make city biking much less nerve-wracking: an AI-powered rear-view digicam that acts just like the collision warning methods already customary in fashionable vehicles. The system, known as Luna Oculus, watches your again and (metaphorically) screams when hazard approaches.
The expertise will not cease vehicles from buzzing too shut or miraculously patch potholes, nevertheless it does provide one thing useful: a number of further seconds to react. In metropolis biking, that may be the distinction that issues.
“Our product will deal with the rider and letting them know when they’re at elevated danger,” explains Andrew Fleury, Luna’s CEO, over e mail. “If attainable, we would like the rider to take motion to scale back their danger and attempt to assume a greater street place relative to the scenario.”
Luna Methods
The Luna Oculus mounts in your bike like an everyday rear gentle – as a result of it’s one (which places out 90 lumens). However inside that housing sits a 1080p digicam with six hours of battery life on a single USB-C cost. The system runs AI fashions that may establish vehicles, vehicles, buses, and different bikes approaching from behind. The reside video feed goes straight to your cellphone securely mounted to your handlebar, so you’ll be able to see what’s coming.
When a car will get too shut, the system escalates alerts based mostly on danger stage. You select the way you need to be warned – visible cues, audio pings, or each – and you may alter the gap threshold that triggers an alarm. The purpose is most consciousness with minimal distraction.
One other key characteristic of Luna is that every one the AI processing occurs on the system itself. No cloud dependence means no latency lag when a truck is respiratory down your neck. “Luna is constructed on a brand new class of AI chip that’s particularly designed for low-power purposes,” says Fleury. “Which means that we are able to run AI fashions in a particularly power-efficient manner. Mannequin optimization for such lightweight purposes is basically our core competence as an organization.”
Luna Methods/Indiegogo
Luna Oculus would not simply warn you, it remembers. The system robotically information harmful incidents throughout your journey: shut passes inside 1.5 m (4.9 ft), automobiles tailgating, aggressive maneuvers. It geotags these moments and maps them, constructing an image of which sections of your route are essentially the most harmful.
This mapping characteristic differentiates Luna from opponents just like the Beam RS 1000, which additionally presents AI-enhanced digicam alerts and automatic incident recording however lacks mapping features.
If you happen to’re keen to share your knowledge, Luna can combination this data throughout a number of riders to create collaborative hazard maps – crowdsourced intelligence about the place the roads are most hostile to bikes. “Luna additionally offers incident clips with post-ride mapping (blackspot identification for route security insights),” Fleury notes. “Moreover, we intelligently report all danger incidents that you could be expertise in your journey so that you even have proof in case the worst scenario arises.”
Digicam-based alert methods have apparent issues after darkish. Luna handles this by way of a mixture of enhanced preprocessing and movement detection algorithms. In low gentle – daybreak, nightfall, well-lit night time scenes – the system maintains object detection functionality.
Luna Methods
In line with Fleury, in very darkish or pitch-black situations, the system nonetheless catches vehicles approaching because of their headlights, although it loses the flexibility to establish unlit objects like pedestrians or different cyclists.
“We at all times maximize the detections which can be attainable within the situations,” says Luna’s CEO. Rain and dust are managed by way of the digicam housing design, and if the lens will get obscured, the system prompts you to scrub it.
All of which sounds promising, however the tech is not delivery fairly but. Luna launched an Indiegogo crowdfunding campaign in 2024 the place the anticipated retail value was given as €199 (about US$235), together with a complimentary 12 months of app entry – after that it is projected to value €72 per yr to make use of the service. “Client shipments are focused for mid‑to-late 2026 and we’ll be protecting everybody up to date on timing and in addition pricing by way of our official channels within the coming months,” Fleury confirms.
The corporate raised €1.5 million in seed funding final month and goals to hit €5 million in complete by the top of 2026. For now, Luna is concentrated on bikes, however Fleury hints at potential motorbike purposes down the street. “We might also provide an aftermarket system for bikes sooner or later, however for now we deal with OEM and Tier 1 tasks.”
Luna Oculus – Smartphone app + linked Wifi digicam
Supply: Luna Systems
