The challenge of safely navigating in complex environments has stalled wider deployment of autonomous robots, according to Stereolabs Inc. The Paris-based company yesterday launched the ZED-X, which it claimed is the first stereo camera designed for robotic navigation and process automation in tough indoor and outdoor environments.
“Today’s robots need to navigate in harsh environments and respond quickly as they operate,” explained Cecile Schmollgruber, CEO of Stereolabs. “Building an affordable, industrial-grade surround 3D perception solution is critical to production-scale deployment of next-generation robotics and smart analytics.”
“Our camera-based solution dramatically simplifies 360-degree spatial perception and is backed by an ecosystem of tools to integrate and control them at a price point that makes it easy to add 3D vision to any machine,” she said.
Stereolabs uses sensor fusion
Traditional navigation systems rely on expensive arrays of lidar sensors and cameras, which suffer from limited fields of view, low resolution, and difficulty recognizing objects, said Stereolabs. The company asserted that it is giving robots artificial intelligence, camera-based surround vision.
By fusing 3D spatial data from multiple ZED-X stereo cameras, Stereolabs said its perception platform allows robots to navigate and detect obstacles in any complex environment. They are also more affordable and easier to install than lidar-based alternatives, it said.
With support from Stereolabs’ software ecosystem, the ZED-X allows rapid commercial deployments of robots in agriculture, construction, logistics, and last-mile delivery, said the company.
ZED-X designed for durability
With its IP66-rated aluminum enclosure, GMSL2 connection, and native multi-camera synchronization, the ZED-X is built for production-scale deployments of autonomous robots in rugged conditions, said Stereolabs.
The ZED SDK and ZED Hub multi-camera control platform provide advanced capabilities for 3D perception, including 360-degree depth sensing, localization, object detection, and skeleton tracking, it added.
The ZED-X camera is suitable for manufacturers ready to mass produce robots for industrial and outdoor environments, Stereolabs said.
Available in two form factors, the ZED-X and the ZED-X Mini, the stereo cameras provide 3D perception at a range of 0.2 to 20 m (0.6 to 65.6 ft.) for navigation and up-close at a range of 0.08 to 12.5 m (0.2 to 41 ft.) for object detection during core process automation. For example, an autonomous tractor could use ZED-X for safe navigation, and ZED-X Mini for crop detection.
Stereolabs also listed the following features:
High-resolution RGB sensor with global shutter
The ZED-X’s 1920x1200 global shutter RGB sensor produces imaging that accurately captures any action-filled environment, said Stereolabs.
Moving scenes are rendered quickly up to 120 fps, it explained. The 3.0 µm pixel size guarantees image quality in both low-light and bright conditions, making the camera suitable for both indoor and outdoor applications.
IMU designed for robots that experience high vibration
The built-in inertial measurement unit (IMU) combines a 16-bit digital triaxial accelerometer and a 16-bit digital triaxial gyroscope to accurately detect motion and measure orientation, said Stereolabs. The low drift ensures that calibration isn’t lost, even on rugged terrain.
Strong vibration robustness allows the IMU can endure harsh industrial environments, said the company. ZED-X can also reduces design costs and offer extended operational lifespan because of its temperature stability, it noted.
A secure GMSL2 connection
Output using the GMSL2 connection protocol supports high-speed video data transfer for robotics. In a multigigabit point-to-point connection, GMSL2 transfers raw video data from the ZED-X to an AI gateway at a speed of up to 6 GB per second.
This means the robot can respond quickly to changes in its environment or fast-moving objects. For large robots, additional cameras can be placed farther from the gateway, at 15 m (49.2 ft.) distance, while still delivering lower latency with less power and a higher frame rate than USB 3.0, without EMI, said Stereolabs.
Optimized for use with NVIDIA’s Jetson AGX Orin supercomputer, the ZED-X can be quickly integrated into an intelligent end-to-end workflow, accelerating deployment from the prototype stage to production-scale implementation, Stereolabs said. Each Jetson module can control four ZED-X cameras, reducing cost, weight, and onboard space requirements.
Multi-camera fusion powered by the ZED ecosystem
In conjunction with the release of the ZED-X, Stereolabs also launched a new multi-camera management platform, ZED Hub, and a new 4.0 version of the ZED software development kit (SDK). This combined offering can provide a 360-degree surround view around a robot, which is necessary for safe navigation.
Users can now fuse data from multiple cameras automatically, eliminating the challenges that previously made this a complex, time-consuming process, said Stereolabs.
The ZED-X is available for preorder at $599, and the ZED-X Mini at $549 per camera. The ZED SDK 4.0 and ZED Hub management software are now available and are compatible with the entire ZED range of cameras.
Stereolabs said its products combine hardware, software, and a cloud platform to provide spatial analytics data and recommendations in real time. By developing deep learning and 3D vision algorithms, the company said it gives machines and physical spaces the ability to see and understand their environment.
With offices in New York and San Francisco, Stereolabs serves more than 100,000 users across industries including transportation, defense, retail, smart city, sports, and manufacturing.
Article topics
Email Sign Up
Related Machine Vision News
