At a press event in San Francisco Tuesday, Stereolabs showed off the real-time 3D SLAM — or simulation localization and mapping — features of its ZED camera, the world’s first high-definition 3D camera for depth-sensing. The camera promises to make creating a detailed three-dimensional map as simple as tossing a drone into the air.
The San Francisco-based startup’s not-so-secret weapon: Jetson TX1. Introduced Tuesday, Jetson TX1 is a credit-card sized module that harnesses the power of machine learning to enable a new generation of smart, autonomous learning machines.
Using NVIDIA’s tools, the Stereolabs team was able to port their code directly from a PC to Jetson TX1. The tools also helped them improve the performance of their 3D mapping software by 50%, allowing them to map 3D environments in real time.
“The Jetson TX1 was a revolution because it nearly eliminated the compromises it took to bring our technology to embedded systems,” said Stereolabs CEO Cecile Schmollgruber.
The result: Stereolabs was able to fly a Jetson TX1-equipped drone packing its Stereolabs ZED 3D camera over an old French chateau to create, in real time, a 3D map of the historic estate as it flew.
3D mapping is just one of a host of next-generation applications for Jetson TX1. We built it to power a new wave of millions of smart devices. Jetson TX1’s advanced GPU allows it to incorporate capabilities such as machine learning, computer vision and navigation.
“In addition to depth sensing … we’re able to run tracking and mapping, in real time, aboard the drone thanks to Jetson TX1 and ZED,” said Stereolabs CTO Edwin Azzam.
The demonstration hints at a host of applications for Stereolabs ZED. With 3D SLAM, a drone or other robot has a detailed understanding of the world around it, meaning implementing autonomous navigation is now easier and accessible to any developer.
Augmented reality is another application: Game developers can use the ZED to create and enhance an imaginary reality. Real estate agents can use it to create immersive digital recreations of their listings. Map makers can use it to chart tough terrain quickly and easily.
“What we’re showing here is how we capture high-quality depths map, track the camera and fuse the 3D reconstructions together to create the 3D mesh of the environment,” said Azzam. “This resulting mesh can be imported into any 3D software, and has really simplified the process of creating and editing a 3D model.”