Vision-based mapping
and road inspection

First scalable solution for continuous
mapping with a simple camera
Your camera is the new LIDAR
Roadly provides a cost-effective and scalable solution that enables fleet-based HD mapping in real time.

We use consumer-grade devices’ built-in monocular cameras instead of expensive sensors. Our proprietary visual SLAM engine precisely positions assets on a map with inch-level accuracy.

Roadly's technology is based on crowdsourced data collection, computer vision, and AI. Our approach requires a minimal level of supervision.
Your camera is the new LIDAR
Roadly provides a cost-effective and scalable solution that enables fleet-based HD mapping in real time.

We use consumer-grade devices’ built-in monocular cameras instead of expensive sensors. Our proprietary visual SLAM engine precisely positions assets on a map with inch-level accuracy.

Roadly's technology is based on crowdsourced data collection, computer vision, and AI. Our approach requires a minimal level of supervision.

Commercial-grade accuracy

with consumer-grade devices

Accuracy and consistency
Our computer vision team's five-year development of our Direct Monocular SLAM engine has culminated in valuable results; we can achieve incredibly accurate dense point clouds, which are constructed with videos from off-the-shelf smartphones.

The stability of our algorithm allows long-distance data collection, which is unachievable for other SLAM engines on the market.

A dense point cloud with 1-inch localization accuracy and mm-level granularity

Color-coded semantic segmentation of a dense point cloud

AI that understands 3D
Using a single camera as a sensor ensures perfect alignment between 2D images and 3D point clouds. Combining visual and spatial data allows unsupervised machine learning algorithms to unlock true potential of AI.

We provide out-of-the-box semantic segmentation with geo-referencing, improved by a spatial understanding of each road feature.
AI that understands 3D
Using a single camera as a sensor ensures perfect alignment between 2D images and 3D point clouds. Combining visual and spatial data allows unsupervised machine learning algorithms to unlock true potential of AI.

We provide out-of-the-box semantic segmentation with geo-referencing, improved by a spatial understanding of each road feature.

Color-coded semantic segmentation of a dense point cloud

A truly scalable approach
The wide availability of smartphones allows for an unprecedented scale of mapping with the lowest possible costs. Each new drive going through our platform improves the final map thanks to our advanced stitching algorithms based on both visual and geometrical principles.

We can run our mapping engine through any device equipped with a camera and rough GPS, from dashcams to iPhones.
A truly scalable approach
The wide availability of smartphones allows for an unprecedented scale of mapping with the lowest possible costs. Each new drive going through our platform improves the final map thanks to our advanced stitching algorithms based on both visual and geometrical principles.

We can run our mapping engine through any device equipped with a camera and rough GPS, from dashcams to iPhones.

Industries

Publications

We use cookies to provide you with the best website experience.
Ok, don't show again