Partnership Accelerates Edge AI Vision Deployment Across NVIDIA Jetson Platforms

AVerMedia Technologies and Stereolabs have rolled out NVIDIA JetPack 7.2 support across all Stereolabs ZED X GMSL cameras and AVerMedia Jetson carrier boards, providing robotics and edge AI developers with an accelerated path from development to production.

The collaboration pairs AVerMedia’s carrier-board hardware and Board Support Package (BSP) engineering expertise with Stereolabs’ unified ZED X driver architecture. Under this unified workflow, AVerMedia adapts the hardware-specific device tree and generates custom drivers for the specific JetPack and Jetson Linux versions required by clients. This approach eliminates the iterative technical exchanges previously needed for every hardware or software update, allowing AVerMedia hardware development and Stereolabs software releases to proceed in parallel. Consequently, engineering teams benefit from shorter validation cycles and greater predictability over deployment timelines.

The structured division of engineering tasks allows Stereolabs, an Ouster company, to concentrate on camera firmware, Image Signal Processor (ISP) performance, and ZED SDK capabilities, while AVerMedia manages carrier-board integration and the validation necessary for commercial deployment.

“By anchoring the platform-specific integration work with AVerMedia, where we’re doubling down on both companies’ hardware expertise, both teams can develop in parallel and respond to customer requirements more quickly,” said Ihsane Debbache, VP of Engineering at Stereolabs. “This integration gives developers a more efficient path to deploying Stereolabs ZED vision across AVerMedia Jetson platforms.”

System memory often serves as a critical operational bottleneck in embedded edge AI architectures, particularly when multi-camera pipelines, computer vision models, and robotic motion workloads must run simultaneously on a single processor module. Addressing this constraint, NVIDIA JetPack 7.2 introduces Jetson agent skills designed to help developers optimize memory allocation across the entire software stack. These agent-driven workflows tune bootloader carveouts, optimize kernel memory reservations, streamline background user-space services, and lower inference memory overhead. Reclaiming this capacity frees essential memory for real-time vision algorithms and complex AI tasks, enabling richer edge applications on smaller hardware footprints while curbing total cost of ownership.

Compatibility extends across AVerMedia’s entire NVIDIA Jetson carrier-board portfolio, supporting both Jetson Orin and Jetson Thor platforms. For production lifecycle management, reproducible Debian packages offer end-to-end traceability across the kernel code, device trees, drivers, and installation flows deployed in the field. Drivers can also be regenerated for subsequent JetPack and Linux releases, streamlining long-term maintenance and system compliance. In addition, NVIDIA JetPack 7.2 provides official Yocto support, granting developers tighter control when building reproducible, production-grade operating system images.

“Customers need both faster integration and more efficient use of their computing resources,” said Alex Liu, Vice President of AVerMedia. “By combining Stereolabs’ vision technology and AVerMedia’s BSP and hardware engineering expertise, we can help customers reduce integration risk, make better use of available Jetson memory, and deploy more capable AI vision systems faster.”

Engineering teams transitioning to NVIDIA JetPack 7.2 can access technical support and ongoing migration resources through AVerMedia’s support form.